Full text - Universiteit Leiden

Transcription

Challenging climate change
Competition and cooperation among pastoralists and
agriculturalists in northern Mesopotamia (c. 3000-1600 bc)
Arne Wossink
Sidestone Press
This publication is a result of the project Settling the steppe. The archaeology of changing
societies in Syro-Palestinian drylands during the Bronze and Iron Ages.
© 2009 A. Wossink
Published by Sidestone Press, Leiden
www.sidestone.com
Sidestone registration number: SSP52410001
ISBN 978-90-8890-031-0
Cover Illustration: The Wadi Jaghjagh as seen from Tell Hamidiyeh.
Photo by the author (mirrored for esthetical purposes)
Cover design: K. Wentink, Sidestone Press
Lay-out: P.C. van Woerdekom, Sidestone Press
Challenging climate change
Competition and cooperation among pastoralists and
agriculturalists in northern Mesopotamia (c. 3000-1600 bc)
Proefschrift
Ter verkrijging van
de graad van Doctor aan de Universiteit van Leiden,
op gezag van Rector Magnificus prof. mr. P.F. van der Heijden,
volgens het besluit van het College voor Promoties
te verdedigen op woensdag 28 oktober 2009
klokke 15:00 uur
door
Arne Wossink
Geboren te Hilversum
in 1978
Promotiecommissie
Promotor: Prof. dr. J.L. Bintliff
Co-promotor: Dr. D.J.W. Meijer
Overige leden: Prof. dr. P.M.M.G. Akkermans
Prof. dr. R.T.J. Cappers
Dr. G. van der Kooij
Prof. dr. E. Peltenburg
Prof. dr. W.H. van Soldt
Contents
List of figures
ix
List of tables
xiii
1 Introduction
1.1 Setting the stage: the context of this study
1.2 The emergence of the late third millennium bc climate change hypothesis
1.3 Studying human responses to climate change: methodological considerations
1.4 The central research question
1.5 About this book
2 Northern Mesopotamia from 3000 to 1600 bc
2.1 Introduction
1
1
2
3
6
7
9
9
2.2 The geographical background
2.2.1 Modern climate
2.2.2 Current vegetation and land-use
9
10
12
2.3 Past environment of northern Mesopotamia
2.3.1 Soreq Cave
2.3.2 Lake Van
2.3.3 Dead Sea
2.3.4 Gulf of Oman and Red Sea
2.3.5 Palaeoclimate records from northern Mesopotamia
2.3.6 Reconstructing climate change
2.3.7 Synthesis of climate change
2.3.8 Evidence for drought
15
17
18
18
19
19
22
24
25
2.4 Chronology construction
26
2.5 Some notes on the political history of northern Mesopotamia
29
3 Theorizing social responses to environmental change
33
3.1 Introduction
33
3.2 Defining the scale of analysis
33
3.3 Ordering social responses to environmental change
34
3.4 Maintaining alternative social responses to environmental change
36
3.5 Modelling social responses to environmental change
3.5.1 Territoriality, social hierarchy, and conflict
3.5.2 Exchange networks and social structure
3.5.3 Hypothesizing social responses to environmental change
37
39
40
42
3.6 Archaeological correlates of competition and cooperation
42
3.7 Conclusions
44
4 Methods for the reconstruction of regional settlement trends
45
4.1 Introduction
45
4.2 Interpreting survey evidence: limitations and possibilities
46
4.3 Problems of phase-based chronology
48
4.4 The contemporaneity problem
4.4.1 Simulating site contemporaneity
4.4.2 Evaluating the site contemporaneity model and its results
4.4.3 Estimating total occupied area
49
50
51
53
4.5 Reconstructing population levels
55
4.6 Rank-size analysis
4.6.1 Inspecting rank-size distributions
4.6.2 Interpreting rank-size distributions
60
60
63
4.7 Conclusions
64
5 Regional settlement trends
65
5.1 Introduction
65
5.2 Selecting the surveys
5.2.1 The Birecik-Euphrates Dam Survey
5.2.2 The Balikh survey
5.2.3 The North Jazira Survey
5.3 Reconstructing regional settlement trends
5.3.1 Demographic trends
5.3.2 Rank-size graphs
65
66
74
81
85
86
89
5.4 Contextualizing the reconstructions
5.4.1 The results from other surveys
5.4.2 The historical evidence
91
91
99
5.5 Conclusions
99
6 The development of pastoralism
101
6.1 Introduction
101
6.2 Defining pastoralism and nomadism
101
6.3 Animal husbandry and pastoralism in pre- and proto-history
102
6.4 Pastoralism during the third and early second millennium bc
104
6.5 The emergence of specialized pastoralism in northern Mesopotamia
6.5.1 Specialized pastoralism during the third millennium bc
6.5.2 Specialized pastoralism during the early second millennium bc
111
111
114
6.6 Conclusions
117
7 The development of social networks
119
7.1 Introduction
119
7.2 The Amorite identity in the textual record
7.2.1 Amorites before the Ur III period
7.2.2 Amorites in the Ur III period
7.2.3 Amorites in the early second millennium bc
119
120
122
125
7.3 Determining the nature of the Amorite identity
7.3.1 Understanding ethnicity
7.3.2 The development of the Amorite identity
7.3.3 The Amorite identity as ethnicity
129
129
131
135
7.4 Conclusions
137
8 Social responses to environmental change
139
8.1 Introduction
139
8.2 The development of sedentary-sedentary relations
8.2.1 The Upper Euphrates: the emergence of Carchemish
8.2.2 The Balikh Valley: cycles of competition
8.2.3 The northeastern Jezirah: centralization and exchange
139
140
141
142
8.3 The development of sedentary-(semi-)nomadic relations
145
8.4 Conclusions
148
8.5 Final comments: looking back and looking into the future
148
Bibliography
151
Samenvatting
177
Dankwoord
181
Curriculum vitae
183
List of figures
Fig. 2.1:
Map of northern Mesopotamia showing major rivers, mountain ranges, modern places, and political borders. Inset map shows the research area in its wider
Near Eastern context.
10
Map of northern Mesopotamia showing mean annual rainfall based on longterm measurements (based on Wirth 1971: Karte 3), and archaeological sites
mentioned in this study.
11
Average monthly rainfall based on a 20-year record (1986–2005) for selected
weather stations in northern Mesopotamia (data from the Syrian Meteorological
Department). (A) Aleppo. (B) Hassakeh. (C) Qamishli. (D) Raqqa.
11
Annual rainfall in hydrological years at selected weather stations in northern Mesopotamia from 1986 to 2005 (data from the Syrian Meteorological
Department). Horizontal line indicates 250 mm. Shading indicates years with
less than 250 mm at all stations. (A) Aleppo. (B) Hassakeh. (C) Qamishli. (D)
Raqqa.
12
Fig. 2.5:
Map of northern Mesopotamia showing simplified potential vegetation zones
under modern climatic regime (based on Moore et al. 2000: fig. 3.7).
13
Fig. 2.6:
Map of the Near East showing palaeoclimate proxy records used in this study
(black stars), as well as sites with evidence for drought in plants and humans
(white stars). (A) Soreq Cave. (B) Lake Van. (C) Dead Sea. (D) Gulf of Oman.
(E) Red Sea. (F) Kazane Höyük. (G) Göbekli Tepe. (H) Wadi Avedji. (I) Wadi
Jaghjagh. (J) Tell Leilan. (K) Sites with botanical samples. (L) Mount Sedom.
(M) Tell Barri.
15
Near Eastern palaeoclimate proxy records used in this study for the period
3100–1500 bc. See main text for references. (A) δ 13C (solid line) and δ18O
(dashed line) values at Soreq Cave. (B) Mg/Ca (solid line) and δ 18O (dashed
line) values, and PAZ (grey dashed line represents transition) at Lake Van.
(C) Dead Sea water level changes. Uncertainties are dashed. (D) Dolomite
(solid line) and CaCO3 (dashed line) values at the Gulf of Oman. (E) δ18O
value at the Red Sea. (F) Environmental history around Kazane Höyük. The
transition is marked by an arrow. (G) δ18O and δ13C values at Göbekli Tepe.
(H) Sedimentation history in the Wadi Avedji. (I) Sedimentation history in
the Wadi Jaghjagh. (J) Stages of the LCCM at Tell Leilan. (K) Averaged Δ13C
values of barley grains from sites in the Levant and northern Mesopotamia.
(L) δ13C values of a subfossil tree from Mount Sedom. (M) Mortuary evidence
from Tell Barri.
16
Annual rainfall in hydrological years for the Jezirah (dark shading, average
from Hasakeh, Qamishli, and Raqqa, cf. Fig. 2.4) and Jerusalem from 1986 to
2005 (light shading, Jerusalem data from the Israel Meteorological Service).
23
Fig. 2.9:
Map of northern Mesopotamia showing late third millennium
sites and regions mentioned in this study.
29
Fig. 2.10:
Map of northern Mesopotamia showing early second millennium bc names of
sites and regions mentioned in this study.
Fig. 2.2:
Fig. 2.3:
Fig. 2.4:
Fig. 2.7:
Fig. 2.8:
bc
names of
32
ix
Fig. 4.1:
Discrepancy between real periods of occupation at a site for hypothetical
phases X, Y, and Z, and the period of occupation as it would be reconstructed
from survey finds from that site.
50
Occupation types a–d for hypothetical phases X, Y, and Z, and transition points
t1 and t2. Letters a–c indicate the period of time during which these site types
were certainly occupied. Letter d indicates a floating period of time in phase
Y during which the site was occupied. Dashed lines indicate a period of time
during which the site may have been occupied (after Dewar 1991: fig. 3).
51
Scattergram indicating the correlation between settlement size and number of
occupants. Crosses indicate the villages drawn from Aurenche 1981, table 3,
and Kramer 1982, table 5.3. Diamonds indicate the towns drawn from Kramer
1982, table 5.6. Triangles represent archaeological sites. Where necessary, the
range of different site size and population estimates is indicated. Number of
occupants and settlement size are given in logarithmic scale.
59
Scattergram indicating the correlation between settlement size and population
density. Crosses indicate the villages drawn from Aurenche 1981, table 3, and
Kramer 1982, table 5.3. Diamonds indicate the towns drawn from Kramer
1982, table 5.6. Triangles represent archaeological sites. Where necessary,
the range of possible population densities resulting from different site size
and population estimates is indicated. Settlement size is given in logarithmic
scale.
59
Fig. 4.5:
Rank-size plot showing (A) log-normal, (B) primate, (C) convex, and (D) primo-convex distributions.
60
Fig. 4.6:
Values for (A) convex distribution, (B) primate distribution, and (C) primoconvex distribution. Positive and negative components of A value have dark
and light shading, respectively (based on Drennan and Peterson 2004: fig. 2).
62
Map of northern Mesopotamia showing spatial coverage achieved by recent
surveys. (A) Birecik-Euphrates Dam Survey (B-EDS). (B) Tishrin Dam surveys. (C) Tabqa Dam surveys. (D) Land of Carchemish survey. (E) Westgazira
survey. (F) Balikh-Euphrates uplands survey. (G) Middle Euphrates survey.
(H) Balikh Survey (BS) and Wadi Qaramogh survey. (I) Harran survey. (J)
Jebel Bishri survey. (K) Wadi Hammar survey. (L) Upper Khabur Survey. (M)
Tell Beydar Survey (TBS). (N) Brak sustaining area survey. (O) 1988 Tell Brak
Survey. (P) Leilan Regional Survey (LRS). (Q) Tell Hamoukar Survey (THS).
(R) Northeast Syria survey. (S) North Jazira Survey (NJS). (T) Lower and middle Khabur survey. (U) Khabur Basin Project. (V) Wadi Ağiğ survey.
66
Fig. 5.2:
Total number of sites (bars) and aggregate site size in hectares (line) for every
B-EDS period without correction for site contemporaneity.
73
Fig. 5.3:
Map of the B-EDS region showing all sites with an Early EBA, Middle EBA,
Late EBA/Early MBA, or Late MBA occupation (based on Algaze et al. 1994:
fig. 2).
73
Fig. 5.4:
BS period without correction for site contemporaneity.
77
Fig. 5.5:
Map of the BS region showing all sites with a Balikh VIab, VIcd, VIIa, VIIb,
or VIIc occupation.
80
Fig. 5.6:
NJS period without correction for site contemporaneity.
84
Fig. 4.2:
Fig. 4.3:
Fig. 4.4:
Fig. 5.1:
Fig. 5.7:
Map of the NJS region showing all sites with a Ninevite 5, Late third millennium, or Khabur occupation (after Wilkinson and Tucker 1995).
84
Fig. 5.8:
Distribution of site sizes (ha) across all periods for the combined BS, B-EDS,
and NJS data sets (n=368).
85
Fig. 5.9:
Settlement trends through time. Solid lines indicate the corrected aggregate
site size, shaded areas indicate the possible range of aggregate site size, bars
indicate the corrected number of contemporaneously occupied sites, and error ranges indicate the possible range of contemporaneously occupied sites.
(A) B-EDS region. (B) BS region. (C) NJS region.
86
Reconstructed population levels and population growth through time. Shaded
area indicates the population size under a population density of 150–250 persons/ha. Dashed line indicates population growth rates through time, based
on corrected site counts and site size estimates. Solid line indicates the combined aggregate site size. (A) B-EDS region. (B) BS region. (C) NJS region.
(D) The B-EDS, BS and NJS regions combined.
87
Rank-size plots for the B-EDS region at (A) 2500 bc, (B) 2200 bc, and (C)
1900 bc, based on corrected site counts and site size estimates. Dashed line is
log-normal distribution. Grey zone represents 90 % confidence range.
89
Rank-size plots for the BS region at (A) 2400 bc, (B) 2000 bc, (C) 1800 bc,
(D) 1700 bc, and (E) 1550 bc, based on corrected site counts and site size estimates. Dashed line is log-normal distribution. Grey zone represents 90 %
confidence range.
90
Rank-size plots for the NJS region at (A) 3000 bc, (B) 2500 bc, (C) 2000 bc,
and (D) 1500 bc, based on corrected site counts and site size estimates. Dashed
line is log-normal distribution. Grey zone represents 90 % confidence range.
91
Uncorrected settlement trends in survey regions in northern Mesopotamia for
the period 3100–1500 bc. Shading indicates relative intensity of settlement.
(A) Birecik-Euphrates Dam Survey (B-EDS). (B) Tishrin Dam surveys. (C)
Tabqa Dam surveys. (D) Land of Carchemish survey. (E) Westgazira survey.
(F) Balikh-Euphrates uplands survey. (G) Middle Euphrates survey. (H) Balikh
Survey (BS). (I) Harran survey. (J) Jebel Bishri survey. (K) Wadi Hammar survey. (L) Upper Khabur Survey. (M) Tell Beydar Survey (TBS). (N) Brak sustaining area survey. (O) 1988 Tell Brak Survey. (P) Leilan Regional Survey
(LRS). (Q) Tell Hamoukar Survey (THS). (R) Northeast Syria survey. (S)
North Jazira Survey (NJS). (T) Lower and middle Khabur survey. (U) Khabur
Basin Project. (V) Wadi Ağiğ survey.
98
Map of northern Mesopotamia indicating sites with faunal assemblages discussed in this study. The rectangles indicate the site clusters used in Figs. 6.2
and 6.4. (A) Turkish Euphrates. (B) Syrian Upper Euphrates. (C) Balikh/Wadi
Hammar. (D) Upper Khabur. (E) Middle Khabur.
105
Fig. 5.10:
Fig. 5.11:
Fig. 5.12:
Fig. 5.13:
Fig. 5.14:
Fig. 6.1:
xi
Fig. 6.2:
Presence of sheep/goat, other domestic, domestic/wild, and wild animals at
fifth to second millennium bc sites in northern Mesopotamia (after Clason and
Buitenhuis 1978; Stein 1987; Wattenmaker 1987; Boessneck 1988; Buitenhuis
1988; von den Driesch 1993; Vila 1995; Stein et al. 1996; Rothman et al. 1998;
Zeder 1998; van Neer and Decupere 2000; IJzereef 2001; Frangipane et al.
2004; Koliński and Piątkowska-Małecka 2008; Buitenhuis forthcoming). Note
that in some cases, the class domestic/wild could not be distinguished on the
basis of the available data. (A) Turkish Euphrates. (B) Syrian Upper Euphrates.
(C) Balikh/Wadi Hammar. (D) Upper Khabur. (E) Middle Khabur.
106
Fig. 6.3:
Map of northern Mesopotamia showing the zone where Kranzhügel are found
(based on Moortgat-Correns 1972: map 2; Lyonnet 1998: fig. 4).
109
Fig. 6.4:
Proportion of sheep/goat in the total assemblage of domestic animals at fifth
to second millennium bc sites in northern Mesopotamia. Sites are chronologically ordered. (A) Turkish Euphrates. (B) Syrian Upper Euphrates. (C) Balikh/
Wadi Hammar. (D) Upper Khabur. (E) Middle Khabur.
119
xii
List of tables
Table 2.1:
Ethnohistorical correlation between various agricultural crops and annual
rainfall (+: significant; +/-: present; -: insignificant) (based on Wilkinson
2004: fig. 3.3).
14
Comparison of various properties of Near Eastern palaeoclimate proxy
records used in this study (A–J), as well as sites with evidence for drought
(K–M). Letters in the left column refer to Figs. 2.6 and 2.7. See main text for
references.
16
Table 2.3:
Summary of the Leilan Climate Change Model (LCCM) as described in Weiss
et al. 1993.
21
Table 2.4:
Comparison of third and early second millennium bc periodizations for the
Jezirah, southern Mesopotamia, selected sites, and the Balikh Survey, BirecikEuphrates Dam Survey, and North Jazira Survey (based on Brinkman 1977;
Curvers 1991; Algaze et al. 1994; Wilkinson and Tucker 1995; Lebeau 2000a;
Dohmann-Pfälzner and Pfälzner 2001; Oates et al. 2001; Pruß 2004; Ristvet
2005; Sallaberger 2007).
28
Table 3.1:
Expected social structures regulating access to resources under different
conditions of resource availability (based on Kelly 1995: fig. 5-6).
38
Table 4.1:
Settlement size (ha), number of occupants per settlement, and population
density for selected excavated sites in greater Mesopotamia dating to the
third and second millennium bc.
59
Survey strategy details for the BS, B-EDS, and NJS. Number of sites refers
to the total number of surveyed sites, with number of sites with evidence
for the period discussed here in parentheses. Letters in the left column refer
to Fig. 5.1.
66
Table 5.2:
Occurrence of ‘fruit stand’ stems and corrugated Hama goblets on sites in
the B-EDS region (Peltenburg 2007b: table 1).
68
Table 5.3:
Size estimates (ha) for B-EDS sites with an EB–MB component (based on
Algaze et al. 1994: with modifications). Note that no site number was assigned to Şavi Höyük II by the B-EDS.
72
Third and second millennium bc chronological concordance for sites surveyed by the BS and Córdoba. Site numbers in the second and third columns
refer to those published in Curvers 1991 and Córdoba 1988, respectively.
75
Table 5.5:
Comparison of widely varying site size estimates by Curvers and Lyon for
Balikh period VII.
76
Table 5.6:
Size estimates (ha) for BS sites with a third millennium bc (Balikh VIa–d) or
early second millennium bc (Balikh VIIa–c) component (based on Curvers
1991: with modifications).
78
Table 2.2:
Table 5.1:
Table 5.4:
xiii
Table 5.7:
Size estimates (ha) for NJS sites with a third–early second millennium bc
component (based on Wilkinson and Tucker 1995: and Ball et al. 1989 (Tell
al-Hawa), with modifications). Second column shows concordance with RJP
site numbers as listed by the Jazirah Salvage Project (Altaweel 2007: table
1).
82
Table 5.8:
Site size classes, number of sites per class, and mean size per class in hectares
(n=368).
85
Table 6.1:
Comparison of available evidence for third and early second millennium bc
pastoral subsistence strategies (+: strong evidence; +/-: some evidence; -:
very little/no evidence).
111
Table 7.1:
Selected textual references to Amorites in chronological order. See main text
for references.
120
Table 7.2:
Comparison of available evidence for ethnic properties of Amorite identity through time (+: strong evidence; +/-: some evidence; -: very little/no
evidence).
131
xiv
1 Introduction
1.1 Setting the stage: the context of this study
This study explores human social responses to environmental change in northern Mesopotamia
during the late third and early second millennium bc, insofar as these can be reconstructed from
the archaeological record. This research developed in two different yet related contexts. The first
context is that of the research project of which this study was a part. The second context is that
of the current interest in future environmental change and its effects on modern society. These
contexts will be discussed below.
This study was part of the research project Settling the steppe. The archaeology of changing societies
in Syro-Palestinian drylands during the Bronze and Iron Ages. The Settling the steppe project was initiated to address common interests of by two earlier Leiden University projects: the excavations of
Tell Deir ‘Alla (Jordan), and Tell Hammam et-Turkman (Syria). A relatively rapid succession of
occupation (re-)establishments and abandonments was observed at Iron Age Tell Deir ‘Alla in the
Jordan Valley, leading to questions about the stability of these occupations and the nature of past
human-environment interaction in the region (van der Kooij 2001). Evidence from the Early and
Middle Bronze Age layers at Tell Hammam et-Turkman in the western Syrian Jezirah suggested
that the site may have played a key role in the relations between urban and rural populations, which
were argued to have been a vital part of life in such marginal environments (Meijer 2000, 2007,
forthcoming). The fact that today both the Jordan Valley and the Syrian Jezirah are important
agricultural production areas made questions about the sustainability and nature of past occupations even more relevant. There was thus a shared interest in human-environment relations in
marginal Near Eastern environments, and the different research lines were brought together in the
Settling the steppe project. Three central research questions emerged from the common interests of
the Deir ‘Alla and Hammam et-Turkman projects emerged: (a) what was the reason for and nature
of human occupation in the steppe zone, (b) how stable was this occupation, and which variables
determined this level of stability, and (c) how did human occupation in the steppe zone relate to
more favourable zones? These general questions were to be applied to the project’s two regional
clusters, focusing on Tell Deir ‘Alla and Tell Hammam et-Turkman respectively, and were to be
specified as necessitated by local environmental, socio-cultural, and historical conditions. The
present study deals with the so-called Hammam et-Turkman cluster. Understanding how human
societies in marginal environments coped with environmental change is critical to answering these
questions, especially in areas where societies had to maintain a careful balance with their environment in order to survive.
The present study furthermore emerged from the recent concerns with climate change, as elaborated in the reports of the Intergovernmental Panel on Climate Change (IPCC). It has now been
established beyond doubt that the world is experiencing global climate change that is at least partly
anthropogenic in nature (IPCC 2007). As a result, modern society faces the tremendous challenge
The Settling the steppe project was funded by the Netherlands Organisation for Scientific Research (NWO; project
number 360-62-020) and based at Leiden University. The project team consisted of Dr. Gerrit van der Kooij (director),
Dr. Diederik Meijer (co-director), Dr. Lucas Petit (post-doctoral researcher, Deir ‘Alla cluster), Eva Kaptijn (PhDstudent, Deir ‘Alla cluster), Dr. Fouad Hourani (geomorphologist, Deir ‘Alla cluster), Ellis Grootveld (archaeobotanist,
Deir ‘Alla cluster), and the present author (PhD-student, Hammam et-Turkman cluster). See Kaptijn et al. 2005 for a
general introduction to the Deir ‘Alla cluster, and Kaptijn 2009 and Petit forthcoming for detailed studies focusing on
that region.
Steppe is not by definition a marginal environment, but only under specific socio-cultural and economic conditions,
because marginality is an expression of the ability of human societies to survive in a given environment. The term
steppe should therefore be strictly used in a descriptive sense and not as a synonym for marginality.
Challenging Climate Change
to adapt to a rapidly changing environment. How modern society should mitigate, and adapt to,
the effects of climate change is subject to worldwide scientific discussion. Archaeological and
historical studies documenting human responses to environmental change can play a role in this
discussion, because decisions about the future are made with knowledge about the past. For that
reason, many researchers, policy makers and authors have turned to the disciplines of archaeology and history to find models of successful and failed adaptations to environmental change, and
to draw lessons from these cases (e.g. deMenocal 2001; Weiss and Bradley 2001; Diamond 2005;
Costanza et al. 2007).
There are numerous archaeological case-studies documenting environmentally triggered social
disruptions across the globe. Many of these studies fit remarkably well into what McIntosh et al.
(2000a: 6) called the ‘macrosocietal technological-economic approach’, adopted among others by
the Intergovernmental Panel on Climate Change (IPCC) in their reports on current climate change.
This approach is characterized by biases manifested ‘in issues of scaling, in overreliance on economic theory, in little consideration for history, and in misplaced concreteness’ that ‘limit understanding of the range of human responses to climate’ (McIntosh et al. 2000a: 4). Archaeological
case-studies that adopt this approach are therefore of limited usefulness to the discussion on current environmental change. This observation justifies a study of past human responses to environmental change that aims to break away from the ‘macrosocietal technological-economic approach’
characterized above. The present study hopes to achieve this break by explicitly championing a
small-scale social and diachronic perspective, paying close attention to regional social, cultural, and
environmental diversity (cf. Brooks 2006: 44).
1.2 The emergence of the late third millennium
bc
climate change hypothesis
Ever since it was first recognised that the transition from the Early to the Middle Bronze Age represented a period of significant social and cultural change in many parts of the Near East, there
have been attempts to explain the changes through a single overarching model (cf. Marro 2007a:
14). The archaeological record of this transitional period was characterized by changes in settlement patterns, the emergence of new cultural traits, and apparently abrupt settlement destruction
and abandonment. Explanations have ranged from foreign invasions or penetration of new (ethnic) groups (Kenyon 1966) to system collapse (Butzer 1997; Peltenburg 2000). Based on ancient
Egyptian texts, Bell (1971) proposed that an interruption of the annual Nile floods may have been
responsible for the social disruption of the contemporary First Intermediate Period in Egypt.
However, this hypothesis could not be substantiated by palaeoenvironmental data at the time and
was not taken up for other parts of the Near East until the early 1990s. In the early 1990s, climate change was put forward as an explanation for the changes observed
in settlement patterns and in the political constellation of the ancient Near East and beyond (Weiss
et al. 1993). The study by Weiss et al. held that (a) soil micromorphological studies at and around
Tell Leilan indicated the occurrence of an abrupt dry phase from 2200–1900 bc, possibly caused
by an otherwise unidentified volcanic eruption, (b) this evidence was supported by indications for
drought elsewhere, notably Egypt, (c) this drought led to (i) the abandonment of urban settlements
in the area around Tell Leilan including Tell Leilan itself, (ii) an overall decline of settled population in the same area, (iii) an increase in pastoral subsistence strategies at the cost of agricultural
production, (iv) the collapse of the Akkadian state due to grain shortages, and (v) the subsequent
collapse of the Ur III state due to pressure from refugees coming from the north and grain shortages as reported in the royal correspondence from Ur, and (d) this collapse was synchronous with
collapses elsewhere in the Old World, notably the Levant, Egypt, the Aegean, and the Indus Valley
(Weiss et al. 1993). In the present study, this model of climate-induced societal change will be
referred to as the Leilan Climate Change Model (LCCM) (see Section 2.3.5 for further
discussion).
Advocating climate change as an explanation for socio-cultural change goes back much further in (Near Eastern)
archaeology however (see Section 1.3, and summaries of previous research in Issar and Zohar 2004; Gronenborn
2005; Rosen 2007).
Introduction
The LCCM has been critically received. Butzer succinctly summarized the criticisms as follows:
‘(1) Are the empirical climatic data valid? (2) If so, was the climatic anomaly effective and synchronous across the area in question? and (3) Did sociopolitical unrest or collapse everywhere follow
upon climatic change, in a form consonant with greater aridity?’ (1997: 249). Indeed, many discussions of the LCCM focused on exactly these points. Firstly, environmental data did not seem to
uniformly support abrupt late third millennium drought (e.g. Bottema 1997). Secondly, historical
sources were uncritically used, and it has not been proven that the Upper Khabur was of critical importance for grain production in the Akkadian state (Zettler 2003). Thirdly, urban collapse
in northern Mesopotamia was probably not as wide-spread as argued by Weiss et al. (DohmannPfälzner and Pfälzner 2001; Oates and Oates 2001; Wossink forthcoming). Fourthly, the cause,
date and nature of the abrupt arid phase are still contested (e.g. Courty 2001; Kuzucuoğlu 2007a).
Fifthly, the role of late third millennium bc environmental stress in other so-called collapses across
Asia is still debated (e.g. Madella and Fuller 2006). Still others argued that the urban collapses did
not result from abrupt climate change, but from frequently recurring strong oscillations in rainfall.
In this scenario, climate change may have taken place but it only marginally affected processes that
were already in operation (Wilkinson 1994, 1997). Finally, inconsistencies in the geomorphological
data from Tell Leilan and a problematic correlation with other palaeoclimate proxy records may
have contributed to the continuation of the debate (see Section 2.3.5).
Nevertheless, the LCCM has continued to be propagated in a relatively unmodified form since
its earliest conception (e.g. Courty and Weiss 1997; Weiss 2000; Ristvet and Weiss 2005; Staubwasser
and Weiss 2006). There is now a stronger focus on globally recognized climate mechanisms, rather
than volcanism, as the cause of the so-called 4.2 ka bp event. Furthermore, the causal link between
abrupt drought and the end of the Akkadian state has been downplayed in favour of an emphasis on the effects of drought observed at and around Tell Leilan itself. These possibilities were,
however, already presented in the 1993 publication. Therefore, they do not represent significant
changes to the LCCM (see Section 2.3.5).
At the same time, many researchers are aware that the basic argument of the LCCM, namely
that there was some form of aridification, has stood the test of time. The shift from a relatively
wet climate during the early third millennium bc to a drier climate similar to that of today during the early second millennium bc is now relatively well documented as a result of the increased
number, reliability, and resolution of Near Eastern palaeoclimate proxy records. Although there is
still conflicting evidence, it seems beyond doubt that, at least in northern Mesopotamia, the third
millennium climate change involved a shift toward more arid conditions (Kuzucuoğlu 2007a).
Even though specific details may still be a matter of debate, the LCCM has thus succeeded in
(re-)introducing climate change as an important variable of socio-cultural change (Akkermans and
Schwartz 2003: 283).
1.3 Studying human responses to climate change: methodological considerations
Before methodology can be discussed, important terms must be defined and clarified. The online
Oxford English Dictionary defines the environment as ‘The conditions under which any person or thing
lives or is developed; the sum-total of influences which modify and determine the development of
life or character’. Halstead and O’Shea (1989a: 2) give a similar definition in which environment is
defined as all external factors acting upon the social unit that is studied. Climate, then, is one of
these external factors, and changes therein are at the same time environmental changes. Humans
outside the studied social unit are strictly speaking also part of that unit’s environment; this is a
matter of the scale at which the study is carried out. Climate ‘in a narrow sense is usually defined
as the average weather, or more rigorously, as the statistical description in terms of the mean and
variability of relevant quantities over a period of time ranging from months to thousands or millions of years’ (IPCC 2007: 78). This definition allows the distinction to be made between climate
variability and climate change. Climate variability represents ‘variations in the mean state and other
statistics (such as standard deviations, the occurrence of extremes, etc.) of the climate on all spatial and temporal scales beyond that of individual weather events’ (IPCC 2007: 79), whereas climate
change is a ‘change in the state of the climate that can be identified (e.g. by using statistical tests) by
changes in the mean and/or variability of its properties, and that persists for an extended period,
typically decades or longer’ (IPCC 2007: 78). Derived from this definition, abrupt climate change can
be understood as climate change where the involved ‘time scales [are] faster than the typical time
scale of the responsible forcing [mechanism]’ (IPCC 2007: 76). Weather is nothing more than
day to day variations in climatic variables within the long-term mean. The difference between the
terms climate change, climate variability, and weather as used here, then, is largely a matter of scale
(Butzer 1982: 23). A further difference between variability and change is that variability is predictable because it is to a certain degree cyclical in nature, whereas change can be conceptualised as
directional, by moving away from previously established conditions toward new conditions. Archaeological and historical records provide a unique and effective means of following chan
ges through time in the demography and social organization of human societies. At the same time,
the past two decades have seen a continuous refinement of palaeoclimate proxy records, and palaeoclimate models. Synchronization of material culture chronologies with palaeoclimate histories
is of prime importance to researchers who wish to evaluate the impact of climate on socio-cultural
change. It is only through correct synchronization that archaeologists can perceive potential linkages between climate and socio-cultural changes.
However, synchronization of palaeoclimate developments and socio-cultural histories, as
reflected in material culture change, is only the first step that must be taken in order to understand
the relation between these two mechanisms. After this crucial first step, it must be demonstrated
that climate was the key factor in determining socio-cultural change. In order to achieve this goal,
models of human-environment interaction must be more carefully considered (Minnis 1996: 57;
Coombes and Barber 2005: 305). Failure to consider the complex interactions between humans
and their environment will lead either to an environmentally deterministic view of socio-cultural
change or to a complete neglect of possible environmental impact on human action and history.
Either of these two approaches has its own drawbacks, as will be discussed below.
Environmental determinism, or ‘the idea that a society’s physical environment can control its
cultural development’ (Coombes and Barber 2005: 303) has long been a strong paradigm in anthropology and archaeology alike. During the second half of the twentieth century, support for environmental determinism waned. The 1990s saw a revival in environmental determinism that may
be associated with concerns about the present anthropogenic climate change, and the realization
that past Holocene climate was less stable than had hitherto been thought. These recent studies
often constructed their argument for climate induced socio-cultural change along three steps: (a)
presenting well-dated high-resolution evidence for climate change in a given region, (b) presenting
archaeological evidence for synchronic change in one or more socio-cultural systems, and (c) pointing out that causality between the observed climatic and socio-cultural change was very likely, given
the synchronicity of these changes. However, insufficient attention was paid to the socio-cultural
mechanisms leading to a given response, and it was not shown that the socio-cultural change was a
response to environmental change at all. In other words, these studies failed to show that climate
change was the critical factor in the observed socio-cultural change (Coombes and Barber 2005:
304–5; Gronenborn 2005: 5). Coombes and Barber called this approach a ‘basic “black box” determinism, correlating climatic and cultural changes and largely neglecting the processes involved’
(2005: 304), of which they argued that ‘little can be said about what forms environmentally triggered transitions might take, why certain cultures should be susceptible to them, or – most pertinently – if the proposed relationships can indeed be corroborated’ (2005: 305). Winterhalder
(1994: 36) similarly argued that there is a general tendency to foreground environmental data,
whereas much more attention should be given to the formulation of models of environmental
In light of this definition, it seems premature to call the 4.2 ka bp event an abrupt climate change, given the fact that
the forcing mechanism of the event is still not understood, and there is still not sufficient agreement on its timing and
nature (see Section 2.3.7).
Even random events that fall within climate variability can to a certain degree be considered cyclical because they
occur at statistically regular intervals (Rosen and Rosen 2001: 536), for instance a spring tide that statistically occurs
once every hundred years.
Introduction
variability that can also be applied in the social sciences (cf. Rappaport 1979: 59–60; Rosen and
Rosen 2001: 539). These problems with current approaches are fundamental issues that any new
study of environmentally triggered social change must tackle.
The reverse of environmental determinism, namely the belief that environmental change cannot cause socio-cultural change, or only in very specific cases, is based on a flawed perception of
human socio-cultural evolution. Early human evolution and prehistory, that is, the emergence of
early modern humans and the adoption or invention of new tool types and technologies including agriculture, are usually studied with reference to the environmental context within which these
developments occurred. This perspective partly results from a relative lack of material relating to
socio-cultural complexity, when compared with later periods. This evolutionary and somewhat environmentally deterministic perspective increasingly tends to make way for a socio-cultural theoretical framework as social complexity and the scale apparent in the material record of the studied
societies increase. Once socio-cultural complexity and diversity become apparent in the archaeological record, explanation of this diversity is given priority, and human-environment relations are
no longer studied. In fact, for later periods these relations are assumed to be rather one-sided, according humans increasing control over their environment as their technology becomes better and
socio-cultural complexity increases (Crumley 1994: 2–3).
This dichotomy between evolutionary and culturally determined histories clarifies the need for
a framework for the study of human-environment relations that allows the recognition and inclusion of multiple causes and causalities. Within this framework, human-environment relations must
be understood as the respective impacts, modifications, adaptations, and influences of humans and
their environments on each other. This definition of human-environment relations supports the
recognition that neither nature, or the environment, nor culture is the determining factor in these
relations. There is a recursive relationship between humans and their environment, and whether
human individuals or groups adapt to or impact on their environment, or do both, must be considered on a case to case basis.
A further important consideration is that human-environment interactions are shaped by past
experiences (Winterhalder 1980: 147). Human individuals and groups base their decisions and actions on what they learned or experienced in the past, and apply this knowledge to the situations
they are currently facing. Thus, these decisions are based on knowledge collected by individuals
and groups (McIntosh et al. 2000a: 24). It can in fact be argued that the environment itself results from an extremely long history of human-environment interaction, especially for the period
and region that are the subject of this study (Köhler-Rollefson and Rollefson 1990; McCorriston
1992). More abstractly, it can be argued that ‘Humans as components of the landscape are not
merely exogenous system disruptors, but one of many contributors to the various environmental
constraints in operation’ (Gragson 1998: 218). In order to fully justify this historicity, any study of
human-environment relations must be explicitly diachronic in nature (Butzer 1982: 279; Crumley
1994: 6; McIntosh et al. 2000a: 8–9; Costanza et al. 2007: 522).
These observations on historicity and the role of humans in the shaping of their environment
lead to the conclusion that the susceptibility of human populations to environmental change is
primarily a social condition. Humans are not inherently susceptible to climate change, but only
as a result of very specific local environmental, socio-cultural, and historical circumstances. As
McIntosh et al. put it: ‘To a human population, an environmental crisis is primarily a matter of the
social realm, implying a failure of adaptation. The truly interesting questions in such cases involve
not just meeting the biophysical challenge but also understanding why human problem solving
seems on critical occasions to become ineffective’ (2000a: 7).
It is therefore suggested that providing evidence for climate change and contemporaneous
socio-cultural change is insufficient to prove causality between these two processes. Rather, it must
be demonstrated that climate change affected human communities in such a way that they could
no longer continue their traditional way of living and had to adapt. In other words, it is not sufficient to prove that the climate became drier during the late third to early second millennium bc,
but it must also be shown that this change resulted in drought, that is, insufficient water to sustain
the biomass of a given region (Barrow 1987: 36). Because drought is a function of environmental
as well as biotic factors, it is in fact a social condition when it is applied to human communities.
As such, drought cannot be postulated from decreasing rainfall levels alone, as drought may also
be triggered by ineffective subsistence strategies or by increasing population levels (Barrow 1987:
36–7). In this sense, northern Mesopotamian palaeoclimate remains a relatively poorly known factor in the present discussion (see Section 2.3.6). It may therefore be in fact much more fruitful to
look not at how societies responded to drought, but to look at which responses could be expected,
had there been drought (see Sections 1.4 and 3.5).
Finally, the scale at which responses to environmental change are studied must be explicitly
considered. Many previous studies found a society or culture to be a convenient or appropriate
unit of analysis (e.g. deMenocal 2001; Weiss and Bradley 2001; Diamond 2005). However, there
are serious objections to the adoption of culture or society as a unit of analysis in the study of
responses to environmental change. Cultures are not ecological units in the sense of human individuals, or populations. If human populations are studied in an ecosystemic context, culture is an
important property of those populations in the maintenance of human-environment relationships,
rather than a distinct unit in and of itself (Rappaport 1979: 62). Societies on the other hand are
complex entities consisting of multiple components, each of which may or may not respond in a
certain way to environmental change. Decisions on how to respond to environmental change are
made not by societies as a whole, but by groups such as communities, households, and ultimately
individuals (McIntosh et al. 2000a: 4). Nevertheless, at each of these levels, decision-making may
be influenced by higher-order groups or individuals, such as elites (Rosen 2007: 147). Recognizing
this diversity, or explaining its absence, is necessary in order to improve our understanding of past
societies. It should therefore be of prime importance to anyone studying the effects of environmental change on human societies to reconstruct the decision-making processes at these diverse
scales, and to identify an appropriate scale for responses to environmental change (see Section
3.2).
1.4 The central research question
As has been outlined above, the present study is explicitly concerned with social responses to environmental change. Rather than focusing on changes in subsistence strategies, the present study
aims to explore changes resulting from environmental change that occurred in the social relations
between groups such as communities and households. Because humans are social beings, relations
with other groups and individuals are fundamental to the maintenance of health and well-being.
When these properties are compromised, changes in these relations may be expected as part of
the adaptation process.
Within the context of the Settling the steppe project, the present study is concerned with the appearance, stability, and abandonment of settlements in the marginal areas of northern Mesopotamia
during the Early and Middle Bronze Ages (c. 3000–1600 bc). As the late third millennium bc was
a period of significant environmental change (see Section 2.3.7), changing social relations may be
expected for these communities. However, these sedentary communities were only part of the
total society living in this region, and they not only had to maintain relations with other sedentary
communities, but also with more mobile groups alluded to by historical records from at least the
late third millennium onward (see Chapters 6 and 7). The central research question thus becomes:
how did social relations develop between sedentary groups, and between sedentary and (semi-)
nomadic groups in response to natural environmental stress in northern Mesopotamia during the
late third to early second millennium bc?
Security maximization against resource stress or shortage, whether resulting from adverse environmental conditions, enemies, or otherwise, is a strong motivation for human behaviour (Halstead
and O’Shea 1989a: 1; Binford 2001: 41). In times of scarcity, much attention and energy will be
Note a similar comment by Lewis on the transitional zone between the desert and the sown: ‘The limits of the intermediate zone were not fixed; they depended on politics more than on geography, particularly on the eastern border,
because the desert “is an economic rather than a geographic expression”. The desert was not necessarily arid and
uncultivable, but rather the area in which the nomads wandered and which was devastated by their flocks and herds’
(1987: 23).
Introduction
devoted to the maintenance of the individual’s and group’s well-being, including the development
and/or modification of social relations with other individuals and groups. These social relations
can be classified under one of two general behavioural strategies to ensure survival for the individual or group: cooperation and competition. The social expressions of these mechanisms are
very diverse and depend on local socio-cultural and ecological/environmental conditions. Thus,
trade and the formation of shared identities are cooperative strategies, whereas territoriality and
warfare are competitive strategies. Whether groups develop cooperative or competitive strategies,
depends on the environmental constraints operating on the involved groups, and on the sociocultural context within which they operate (Section 3.5).
Detailed discussion of these concepts will be deferred to Chapter 3 but the resulting hypotheses will be stated in advance here. Application of the concepts of competition and cooperation
leads to the formulation of two hypotheses regarding the development of social relations in northern Mesopotamia during the late third-early second millennium bc: (a) it is to be expected that, as
a result of environmental stress, social relations between sedentary agricultural communities will
develop along competitive lines in order to improve each communities’ chance of survival and access to scarce and therefore contested resources, and (b) sedentary and (semi-)nomadic groups will
develop cooperative mechanisms ensuring mutual survival and maximization of their respective
subsistence strategies.
1.5 About this book
This book is divided in eight chapters, including this introduction. Chapter 2 provides the geographical, environmental and historical background of the study area and discusses the evidence
for climate change during the third and early second millennium bc. Additionally, it considers the
archaeological and historical framework, and the political history of the region.
Chapter 3 explores the theoretical foundations of the previously advanced hypotheses. It furthermore aims to highlight archaeological correlates for the various responses. Rather than rely on
the patchy historical record to infer conflicts between settlements and (city-)states, the development and organization of settlement patterns is presented as a suitable method to follow changes
in relationships between sedentary communities. The development of supra-regional social identities is presented as a social mechanism that can strengthen ties between sedentary and (semi-)nomadic communities. It is argued that changes in these two mechanisms correspond to changes that
would be expected in the face of environmental stress.
Chapters 4 and 5 are concerned with the development of social relations among sedentary
communities as can be gleaned from the record of excavated and surveyed sites. Because survey
data do not allow periods of settlement use and abandonment to be precisely established, it is
necessary to determine the degree to which settlements dating to the same chronological period
were actually contemporaneous. Chapter 4 presents a method for estimating the number of contemporaneous settlements, and discusses approaches to determining the number of persons living
at a site in order to establish community sizes and changes therein. In Chapter 5, three regions are
selected for which sufficient data are available from surveys and/or excavations. After discussing
these case-studies and detailing the steps that are taken in constructing the databases, the statistical
analyses outlined in Chapter 4 are performed on each dataset, and the results are compared.
Chapter 6 explores the organization of pastoralism in northern Mesopotamia during the third
and early second millennium bc. Historical and archaeozoological data are the primary sources for
this analysis. Although these records are unevenly distributed over the period and region under
discussion, it is nevertheless possible to postulate the emergence of a specialized (semi-)nomadic
pastoralism during the late third to early second millennium bc.
Chapter 7 is concerned with changes in the social relations between sedentary and (semi-)
nomadic communities. The reconstruction of these relations draws primarily on historical sources,
since other archaeological evidence pertaining to third and second millennium bc (semi-)nomadic
groups is scarce, if not absent. Even so, it must be realized that the historical sources also tend
to present a one-sided view, since these texts were primarily written by city-dwellers. However,
despite this bias, it will be argued that these sources may be used in a reconstruction.
Chapter 8 brings together the evidence presented in Chapters 4 through 7. This chapter evaluates the research question and hypotheses presented above and in Chapter 3, and discusses them
in light of the data presented in the previous chapters. Here, the degree to which the data conform
to the expected patterns is discussed, and explanations for any deviations are offered. Central to
this chapter, and to this study as a whole, is an overview of the social responses to late third to
early second millennium bc climate change in northern Mesopotamia. Some final comments and
directions for further research conclude this chapter and this study.
2 Northern Mesopotamia from 3000 to
1600 bc
2.1 Introduction
This chapter aims to reconstruct the environment and history of the northern Mesopotamian
world during the third and early second millennia bc. It starts with an outline of the geography of
northern Mesopotamia of this study (Section 2.2) in which the current climatic and environmental
conditions and the potential vegetation are discussed. Furthermore, the effects of recent land-use
patterns on the northern Mesopotamian landscape will be evaluated. Section 2.3 presents the available data on past climate changes in northern Mesopotamia. Section 2.4 outlines the archaeological
and historical chronologies that will be employed here. Finally, Section 2.5 outlines the political
history of the region in so much as it can be reconstructed from written records. 2.2 The geographical background
Geographically as well as geologically, modern Syria can be divided into zones that are roughly
aligned along an east-west axis (Wirth 1971: 17). Bordering the Mediterranean Sea is a narrow
coastal strip that until recent times was insalubrious and very sparsely settled. To the east, the land
rises up to the mountain ranges of the Amanus, Jebel Ansariya and Libanon. Continuing to the
east, the land descends into the valleys of the Orontes and the Jordan Rivers, and rises up again
into the lower mountain ranges of the Jebel Samane, Zaouiye, Anti-Libanon and Hermon. Finally,
the large flat steppes of inner Syria are reached, which stretch eastwards to the Zagros Mountains
and are bordered by the Taurus in the north. These flat steppes are intersected by several rivers,
of which the Euphrates and Tigris are the most important ones. Between these two rivers lies the
area that is today known as the Jezirah (Fig. 2.1).
The Jezirah is part of the Arabian plate that slopes down towards the east. The northern Jezirah
consists of flat plains, with few striking relief features. Of these, the Jebel Abd al-Aziz and the
Jebel Sinjar are the most marked, with heights of 920 and 1480 m respectively. Using both rainfall
and geological data, Reifenberg (1952: 86–7) described the soils of the Jezirah for those regions
receiving less than 200 mm of precipitation annually as desert soils, and as Mediterranean steppe
soils and alluvial Mediterranean steppe soils those areas receiving more than 200 mm. The northern half of the Jezirah is characterized by calcareous soils, whereas to the south gypsiferous soils
dominate. In the northeast corner of the Jezirah, reddish-brown loams on igneous rock occur
(Wolfart 1967).
Rivers and wadis in the northern Jezirah have cut into the plains, resulting in relatively shallow
valleys. Most of the rivers and wadis drain into the Euphrates. The Khabur and Balikh are the most
prominent of these rivers. The Khabur is fed by the karstic springs of ‘Ras al-Ain, and is joined by
multiple wadis including the Wadi Jaghjagh and Wadi Radd, creating the Khabur Triangle. Today,
the Wadi Jaghjagh is an intermittent stream but it carried water permanently until quite recently.
The term northern Mesopotamia is used very loosely here to designate the upper part of the Euphrates and Tigris basins
and the land in between, including the foothills of the Taurus Mountains. As such, it is largely synonymous with the
Jezirah, the term currently used to designate the land between the Euphrates and Tigris rivers north of modern Baghdad
and south of the Taurus. The term greater Mesopotamia is cursorily used to describe the Jezirah and the Tigris–Euphrates
alluvium together (cf. Wilkinson 2000a: 222).
Throughout this chapter and those that follow, historical dates will be expressed as normal calendar years ( bc). Where
necessary, radiocarbon ages were calibrated using the calibration programme OxCal 3.10 and are expressed as calibrated
calendar years (cal bc).
Tur Abdin
Turkey
Tig
ris
Mardin
Jarr
ah
J ag
hjag
Khanzir
Kha
bur
Tall Abyad
Sajur
h
Avedji
Qamishli
Khnezir
Sanliurfa
Radd
Hassakeh
Jebel Sinjar
Balik
h
Jebel Abd al-Aziz
La
ke
A
Jabbul
ssa
d
Syria
Raqqa
Euph
ra
ri
l Bish
J eb e
Palmyra
tes
Iraq
Deir ez-Zor
2000+ m
1500-2000 m
1000-1500 m
500-1000 m
0-500 m
0 km
intermittent lake
lake
N
30
60
Fig. 2.1: Map of northern Mesopotamia showing major rivers, mountain ranges, modern places, and political borders. Inset map
shows the research area in its wider Near Eastern context.
Mean discharge of the Khabur was 50 m3/s with a peak of 300 m3/s and a minimum of 35 m3/s.
The Balikh received most of its water from the spring at ‘Ain al-Arus, just south of the present-day
Turkish–Syrian border, and had an average discharge of 6 m3/s, with a minimum flow of 5 m3/s
and a maximum flow of 12 m3/s. By comparison, Euphrates discharge in Syria ranged between
5000 m3/s in April-May, and 250 m3/s in August–December, with a mean of 840 m3/s (Wirth
1971: 109–10). Average discharge of the Tigris near Baghdad was 1236 m3/s, with a maximum of
2909 m3/s in April, and a minimum of 352 m3/s in October (Buringh 1960: 51). Note that most
of these figures describe the situation before the introduction of large-scale irrigation works and
the construction of hydroelectric dams in the Euphrates, Khabur, and Tigris Rivers in Iraq, Syria,
and Turkey.
2.2.1 Modern climate
Climate and geography create distinct local weather patterns. Today, the Jezirah has a dry continental climate characterized by hot, dry summers and cool, wet winters. Winter climate is heavily
influenced by moisture-carrying cyclones from the Atlantic Ocean and the Mediterranean Sea. The
lion’s share of this moisture is deposited in western Syria, where the Jebel Ansariya and Lebanon
Mountains create a rain shadow for the interior. Moisture is partly carried further east and deposited in northeast Syria, where the Taurus–Zagros Mountains act as a barrier (Fig. 2.2). During
summer, winds come from Anatolia or Iran and cause dry, stable weather. Rainfall is limited to fall,
winter, and spring; the summer months being almost completely dry. Temperature and potential
evapotranspiration reach their maxima during summer. In western Syria, December and January are
the wettest months, whereas to the east, including the Jezirah, the highest precipitation is recorded
for January and February (Fig. 2.3). From west to east, reliability of rainfall generally decreases as
10
Northern Mesopotamia from 3000 to 1600 bc
600
1000
mm
mm
Gritille
Hayaz
Kurban
60 0
Hacinebi
Zeytinli Bahçe
Mozan
Leilan
300 mm
Carchemish
Shaqrah
Gudeda RadRaqa'i
Ziyadeh Atij
Bderi Umm Qseir
Banat
Hadidi
Sweyhat
400 mm
Mashnaqa
Ras et-Tell
As-Saman
Umm el-Marra Habuba Kabira
Halawa
Selenkahiye
Brak
Kashkashok II-IV
Kuran
Chuera
Hammam et-Turkman
Hamoukar Al-Hawa
Hamidiyeh
Barri
Beydar
Jidle
Sahlan
Mohammed Diyab
Arbid
Chagar Bazar
El-Qitar
mm
400 mm
Kazane
200 mm
Emar
Bi'a
Al-Rawda
Terqa
2000+ m
1500-2000 m
1000-1500 m
500-1000 m
0-500 m
0 km
intermittent lake
lake
Qraya
10
m
0 m Mari
N
30
60
Fig. 2.2: Map of northern Mesopotamia showing mean annual rainfall based on long-term measurements (based on Wirth 1971:
Karte 3), and archaeological sites mentioned in this study.
80
60
50
50
June
July
May
June
May
August
April
April
July
March
March
August
January
February
July
August
June
May
April
0
March
10
0
February
20
10
January
30
20
December
February
40
30
November
December
50
October
January
60
50
mm
70
60
September
December
80
70
40
October
D
November
July
August
May
June
April
March
January
February
0
December
0
October
10
November
20
10
September
30
20
November
40
30
September
40
October
mm
70
60
80
mm
C
B
70
September
80
mm
A
Fig. 2.3: Average monthly rainfall based on a 20-year record (1986–2005) for selected weather stations in northern Mesopotamia
(data from the Syrian Meteorological Department). (A) Aleppo. (B) Hassakeh. (C) Qamishli. (D) Raqqa.
11
A
800
700
600
mm
500
400
300
200
100
0
800
B
700
600
mm
500
400
300
200
100
0
800
C
700
600
mm
500
400
300
200
100
0
800
D
700
600
mm
500
400
300
200
2004/2005
2003/2004
2002/2003
2001/2002
2000/2001
1999/2000
1998/1999
1997/1998
1996/1997
1995/1996
1994/1995
1993/1994
1992/1993
1991/1992
1990/1991
1989/1990
1988/1989
0
1986/1987
1987/1988
100
Fig. 2.4: Annual rainfall in hydrological years at selected weather stations in northern Mesopotamia from 1986 to 2005 (data
from the Syrian Meteorological Department). Horizontal line indicates 250 mm. Shading indicates years with less than 250 mm
at all stations. (A) Aleppo. (B) Hassakeh. (C) Qamishli. (D) Raqqa.
does the amount of rainfall. Comparison of annual precipitation from selected weather stations in
and around the Jezirah suggests a high correspondence between these locations. Fig. 2.4 indicates
that in three out of twenty years, rainfall at all stations drops below 250 mm. In other words, low
rainfall in one area is likely to be matched by low rainfall elsewhere (Wilkinson 2004: 44). Longterm rainfall and temperature observations seem to indicate that dry years occur in groups, rather
than individually dispersed among average years. Wirth (1971: 92) has calculated that during such
dry years, the 250 mm isohyet coincides with the long-term average 400 mm isohyet. Conversely,
wetter than average years seem to cluster as well (Wirth 1971: 97).
2.2.2 Current vegetation and land-use
The potential vegetation of the Jezirah, that is the vegetation as it would exist in the absence of
human activities such as agriculture, pastoralism, and deforestation, has been reconstructed by
Moore et al. (2000). For this reconstruction, they used a combination of climatic, botanical, geo-
12
graphical, and historical evidence, and the vegetation as it survives in areas that have been relatively
shielded from grazing and human intervention. Eight vegetation zones are distinguished, ranging
from true forests in the mountain ranges along the Mediterranean coast to true desert in Syria’s
interior. For the northern Jezirah, the number of vegetation zones encountered is more limited.
The vegetation zones are roughly aligned on a north-south axis, except for vegetation zones related to river valleys and mountain ranges (Fig. 2.5). For the northern parts of the Jezirah, dense
forest-type vegetation was reconstructed, dominated by various types of oak, including the Turkey
oak (Quercus cerris) and Lebanon oak (Q. libani). Other important plants include various almond
(e.g. Amygdalus communis subsp. microphylla) and terebinth species (e.g. Pistacia atlantica, P. palaestina).
This vegetation type can potentially be supported in areas with annual rainfall exceeding 400 mm.
In slightly drier areas, this oak-dominated forest gives way to park-woodland vegetation. Oak is
still present, but its distribution is less continuous and interspersed with grassland areas. Multiple
shrub species occupy the border zones of the oak patches in these park-woodlands. Still further
the south, as the Jezirah becomes progressively drier, forest trees can no longer be supported,
except for some drought resistant species. Overall tree density is low, and steppe species such as
Artemisia spp., dominate. True steppe appears further south, well below the 200 mm isohyet. This
area is today dominated by Artemisia and Chenopodiaceae, but would support a large variety of
grasses in the absence of heavy grazing. The Euphrates and Tigris valleys, as well as parts of the
Balikh and Khabur, support riverine forest vegetation, including poplar (Populus euphratica), ash
(Fraxinus rotundifolius), and elm (Ulmus spp.). Swamp areas, as can potentially be encountered along
the Euphrates, Balikh, and the Wadi Radd, would support wetland species. The borders between
zones are generally correlated to rainfall levels but plant species could extend well beyond their hy-
dense woodland
open woodland
woodland steppe
medium-dry steppe
desert/steppe 0 km
intermittent lake
lake
N
30
60
Fig. 2.5: Map of northern Mesopotamia showing simplified potential vegetation zones under modern climatic regime (based on
Moore et al. 2000: fig. 3.7).
13
pothesized zones due to the moisture carrying capacities of specific soils. Especially in and around
wadis, and in coarse-grained soils, plant species may be encountered that need more water than
rainfall alone would offer (Moore et al. 2000).
However, the current vegetation in the northern Jezirah, and the Near East in general, is characterized by millennia-long human modification and use and far removed from the potential vegetation described above (Köhler-Rollefson and Rollefson 1990; McCorriston 1992). Like natural
vegetation, cultivated crops are subject to environmental constraints, particularly rainfall, in the
absence of artificial irrigation and other agricultural production-enhancing strategies. Thus, as
rainfall diminishes, drought-tolerant crops will probably increasingly dominate the crop spectrum.
The number of production strategies aimed at reducing the effects of water shortage, such as fallowing, will also increase. In addition, the overall number of crops is likely to decrease with lower
precipitation. Under relatively recent production strategies, as rainfall diminishes from over 300
mm per year to less than 200, wheat is gradually replaced by barley as the main cereal crop, and
the variety of other crops also diminishes significantly (Wilkinson 2004: fig. 3.3). Below 200 mm,
agriculture as a subsistence strategy becomes altogether unimportant. The 200–300 mm isohyet
clearly represents a border zone in terms of the dominant subsistence system, with agriculture being gradually replaced by pastoralism. Even so, specific soil properties in the relatively dry areas
can allow farmers to grow crops with high moisture requirements there (Wilkinson 2004: 43).
Today, the greater part of the northern Jezirah is cultivated, with cereals being the main crop,
followed by cotton (Lévêque n.d.: 11–13). Much of this cultivation is sustained through irrigation,
practised both on a local and regional scale. Artificial wells tap the groundwater table, which as a
result has dropped considerably in recent decades. Extensive canal networks provide water from
the major reservoirs in the Euphrates and Khabur rivers, or are projected to do so in the near future (Hole and Zaitchik 2007: fig. 4). Mechanical pumps distribute water from perennial as well as
intermittent streams to the fields. As a result of the irrigation possibilities, summer cropping in the
northern Jezirah, requiring large amounts of water and previously virtually non-existent, increased
dramatically in recent years. Additionally, there has been a shift in the location of agricultural plots
away from the river valleys towards off-river plots (Hole and Zaitchik 2007: 142). The current agricultural production regime with its heavy reliance on irrigation water is causing problems of soil
salinization and groundwater quality degradation, and agricultural plots and farms in unfavourable
locations have been abandoned in recent years (Hole and Zaitchik 2007: 144).
The currently established practice of mechanized cultivation can have disastrous effects on
local environmental conditions if agricultural fields are abandoned, as has been observed in Iraq
(Thalen 1979). Cultivation requires that natural vegetation is cleared, large stones are removed, and
fields are deeply ploughed. In marginal areas, the most fertile fields will be cultivated first, for instance wadi bottoms or areas with slightly higher rainfall. These areas are also the most productive
grazing lands for pastoralists. A few dry years may lead to abandonment of these fields, and relocation of cultivation to more fertile lands. On the marginal fields, exposed and loosened topsoil
remains, which is easily affected by wind and water erosion. Surface sealing, whereby a hard crust
is formed, or the formation of a desert pavement may inhibit the re-colonization of the area by the
original vegetation. Abandoned fields are likely to be colonized by species with low grazing value.
barley/fallow
wheat
300> mm
300–200 mm
<200 mm
+/-
+
-
+
+/-
-
fodder crops
+/-
+/-
-
grain, legumes, summer crops, tree crops
+/-
-
-
Table 2.1: Ethnohistorical correlation between various agricultural crops and annual rainfall (+: significant; +/-: present;
-: insignificant) (based on Wilkinson 2004: fig. 3.3).
14
One reason why barley is more suitable as subsistence crop in drier areas than wheat is that barley is more salt-tolerant.
According to Barrow, wheat can take water with a maximum salinity of 15000 parts per million (ppm), whereas barley
has a salinity tolerance of 35000 ppm (1987: table 7.4).
Vegetation regeneration would require between five and fifteen years without grazing and changes
in soil condition. This means that short-term unsuccessful cultivation of marginal lands will lead
not only to the destruction of potential farming lands, but also to the degradation of land that is
normally used for grazing flocks of sheep and goat by pastoralists (Thalen 1979: 295–6, 310).
Sheep/goat pastoralism remains important in the Jezirah and the more arid regions south of
the Euphrates (Lewis 1988; d’Hont 2004; Lévêque n.d.: 15, 24–5). Three types of pastoral populations can be distinguished in the ethnohistorical past in Syria: Bedouin who relied on nomadic
camel-rearing, full-time nomadic pastoralists relying on sheep/goat, and semi-nomadic sheep/
goat pastoralists. Nomadic sheep/goat pastoralists traversed between 50 and 200 km and spent
the summer above the 200 mm isohyet. There, the sheep grazed on fallowing fields or wastelands.
During the colder winter months, the pastoralists camped more to the east and south of the summer pastures, in areas that during the summer were often used by Bedouin. Nomadic sheep/goat
pastoralists left the summer pastures in November and stayed in their winter camps until May. The
largest Syrian tribes had winter pastures northwest of the line Damascus–Palmyra–Raqqa. Seminomadic pastoralists (Halbseßhaften) were former fully nomadic pastoralists who had settled during
the French mandate period and undertook transhumance-like activities for only part of the year
(Wirth 1971: 256–8). Hole (1991) described a migration pattern that was dependent on the available precipitation, that is the wetter the summer season, the farther nomadic pastoralists could
move away from the permanent water sources (see Section 6.2 for a full discussion of the terms
pastoralism, (semi-)nomadism, and transhumance).
2.3 Past environment of northern Mesopotamia
This section discusses relevant climate proxy records, the combined results of which will be used
to reconstruct the palaeoclimate of the Jezirah. Focus will be on high-resolution records outside northern Mesopotamia, as well as low-resolution local records (Table 2.2, Figs. 2.6, 2.7). 10
Furthermore, discussion will be limited to the centuries framing the late third millennium climate
change. In other words, focus will be on the period from 3000 to 1500 cal bc. Before local palaeoclimate proxy records are discussed, the evidence from high latitude ice and deep sea cores
N
B
F
K
K
G
H
K J
0 km 200
400
MI
K
K
K
K
A
C
L
E
D
Fig. 2.6: Map of the Near East showing palaeoclimate proxy records used in this study (black stars), as well as sites with evidence
for drought in plants and humans (white stars). (A) Soreq Cave. (B) Lake Van. (C) Dead Sea. (D) Gulf of Oman. (E) Red Sea.
(F) Kazane Höyük. (G) Göbekli Tepe. (H) Wadi Avedji. (I) Wadi Jaghjagh. (J) Tell Leilan. (K) Sites with botanical samples. (L)
Mount Sedom. (M) Tell Barri.
10 Palaeoclimate proxy records documenting late third millennium climate change elsewhere in the Near East are discussed in
Rosen 1995 (Palestine), Bell 1971, Said 1993, Krom et al. 2002 (Nile), van Zeist and Bottema 1977, Stevens et al. 2001, 2006
(Iran), and Eastwood et al. 1999 (Turkey). See Arz et al. 2006 for a discussion of late third millennium climate change in the
monsoon system. See Issar and Zohar 2004, Mayewski et al. 2004, Staubwasser et al. 2006, and Rosen 2007 for extensive
treatments of Near Eastern climate during the entire Holocene.
15
map
record
proxy
resolution
range
dating
A
Soreq Cave
δ18O / δ13C analysis
~50 yrs?
60 ka bp–present
TIMS 230Th/234U dates
B
Lake Van
δ O / trace element analysis
~1 yr
14.7 ka bp–present
varve counting
C
Dead Sea
lake-level curve
10-100 yrs
8 ka bp–present
14
D
Gulf of Oman
δ18O / CaCO3 / dolomite analysis
~100 yrs
200 ka bp–present
AMS 14C-dates
E
Red Sea
δ18O analysis
~200 yrs
5.9 ka bp–3.9 ka bp
AMS 14C-dates
F
Kazane Höyük
geomorphological analysis
low, discontinuous
12 ka bp–present
cultural chronology
G
Göbekli Tepe
δ18O / δ13C analysis
low, discontinuous
10 ka bp–4 ka bp
14
H
Wadi Avedji
low, discontinuous
8 ka bp–present
cultural chronology
I
Wadi Jaghjagh
low, discontinuous
3.5 ka bc–present
TL dates
J
Tell Leilan
low, discontinuous
2.6 ka bc–1.9 ka bc
14
K
Sites with botanical
samples
Δ13C analysis
low, discontinuous
14
L
Mount Sedom
δ13C / δ15N analysis
low, discontinuous
14
M
Tell Barri
dental analysis
low, discontinuous
EB II-Neo–Assyrian
cultural chronology
18
C-dates
C-dates
C-dates, cultural
chronology
C-dates, cultural
chronology
C-dates
E
F
desiccation, seasonally confined rainfall
relatively arid
heavily forested, high water table, regular river discharge and rainfall
maximum humidity, presumably higher temperatures
D
d18O (‰)
d18O (‰)
sea-level (m)CaCO3 (%wt.) d18O (‰)
phase
mortuary evidence indicates enamel hypoplasia,
dental caries, and high infant mortality
4
stage D13C values
-30
-20
-25
15
16
-1.5
0.5
-370
-380
-390
-400
-0.5
0
LC 5
0.5
3100
1
EJ 0
1.5
3000
-6.0
EJ I
-5.5
2900
-5.0
2800
17
EJ II
18
2700
M
1
EJ IIIb
2500
4
2400
EJ IIIa
L
3
5
EJ IV
2600
K
2
EJ V
2200
decreased stream velocity/increased aridity (?)
2100
2300
J
6
4
OJ I
I
vigorous, relatively steady flow,
riverine gallery forest, swamp
8
20
2000
V-7 V-8
1800
OJ II-III
H
0
12
30
0
1700
1900
G
dolomite (% wt.)
50
-12
MJ I
C
40
1600
100
B
Mg/Ca
-11
A
d13C (‰)
-10
date BC Jezirah
-0.5
Table 2.2: Comparison of various properties of Near Eastern palaeoclimate proxy records used in this study (A–J), as well as
sites with evidence for drought (K–M). Letters in the left column refer to Figs. 2.6 and 2.7. See main text for references.
d13C (‰)
Fig. 2.7: Near Eastern palaeoclimate proxy records used in this study for the period 3100–1500 bc. See main text for references.
(A) δ13C (solid line) and δ18O (dashed line) values at Soreq Cave. (B) Mg/Ca (solid line) and δ18O (dashed line) values, and PAZ
(grey dashed line represents transition) at Lake Van. (C) Dead Sea water level changes. Uncertainties are dashed. (D) Dolomite
(solid line) and CaCO3 (dashed line) values at the Gulf of Oman. (E) δ18O value at the Red Sea. (F) Environmental history
around Kazane Höyük. The transition is marked by an arrow. (G) δ18O and δ13C values at Göbekli Tepe. (H) Sedimentation
history in the Wadi Avedji. (I) Sedimentation history in the Wadi Jaghjagh. (J) Stages of the LCCM at Tell Leilan. (K) Averaged
Δ13C values of barley grains from sites in the Levant and northern Mesopotamia. (L) δ 13C values of a subfossil tree from Mount
Sedom. (M) Mortuary evidence from Tell Barri.
16
should be mentioned. It has long been recognized that cold spells in high latitudes may indicate
increased aridity in lower latitudes, as more water is stored in glaciers and ice sheets. For the third
millennium bc, the Greenland Ice Sheet project 2 (GISP2) ice core indicates cold spells, which are
correlated to arid events at lower latitudes, in 2800, 2650, 2450, 2350, 2050, and 1950 bc. Deep sea
cores from the North Atlantic document the occurrence of drift ice, and these cores suggest cold
periods around 2600 bc and 2200–2050 bc (Kuzucuoğlu 2007a: 463–4). It should be noted, however, that there was considerable variety in intensity of these events, and that numerous smaller
changes in temperature occurred during the third and second millennium. Recently, a correlation
has been proposed between these events and magnetic reversals documented in mudbricks from
Mari, suggesting that these high-latitude events indeed affected the Near East in some way (Gallet
et al. 2008).
2.3.1 Soreq Cave
The proxy record from Soreq Cave, Israel, is among the most important of the Near East. It is located at the intersection of climate systems from Europe, Africa, and Asia, and hence susceptible
to changes in any of these. Today, Soreq Cave experiences a climate with an average annual rainfall
of 500 mm, mainly deposited during winter. Water entering the cave deposits annually laminated
calcite speleothems that have been used to construct a continuous record of climate change for
the last 185000 years. Palaeoclimate proxies that are analyzed include δ 18O and δ13C values and uranium and strontium isotope ratios in the speleothems. Chronological control is achieved through
multiple thermal ionization mass spectrometry (TIMS) 230Th/234U dates, taken at regular intervals
(Bar-Matthews et al. 1997; Bar-Matthews et al. 2003).
The oxygen isotope and carbon isotope values are important in the present context. The δ 18O
value indicates the ratio of the 18O isotope to the 16O isotope. Variations in the 18O/16O ratio indicate variations in temperature and/or precipitation, depending on the proxy record that is studied.
For Soreq Cave, it has been shown that δ18O variations can be correlated to changes in precipitation, whereby low δ18O values indicate moist periods, and high δ18O values indicate dry periods
(Bar-Matthews et al. 1997: 161). By applying the current correlation between annual amount of
rainfall, δ18O values of rainfall, and δ18O values of the corresponding speleothems to the entire
speleothem record, it is possible to reconstruct changes in rainfall through time (Bar-Matthews
et al. 1998). The carbon isotope composition of speleothems is influenced by numerous factors,
including the picking up of plant-derived CO2 as rainwater passes through the ground towards the
cave. This process allows δ13C to be used as an indication of the ratio of C3 to C4 plants, which
grow in moderate and drier climatic conditions, respectively. Thus, high δ 13C values generally indicate a stronger presence of C4 plants and by inference drier conditions, whereas low δ 13C values indicate fewer C4 plants and moister conditions. It should be noted, however, that due to the
complexities of speleothem formation, these correlations are very rough approximations. Under
specific environmental conditions, the described processes may have operated in slightly different
ways, and human actions may have played a role as well (Goodfriend 1999: 510).
The δ18O and δ13C values recorded at Soreq Cave suggest the following reconstruction of
climate trends (Fig. 2.7A). For the Holocene, the combined δ 18O and δ13C records clearly document the major climatic shifts of the Late Glacial Maximum (LGM), the Younger Dryas, and the
Holocene dry event at 6100 cal bc. After this event, there followed a wet period during which δ 18O
and δ13C values were generally lower than today. The period from c. 3000 to 1500 cal bc is characterized by several extreme fluctuations in δ18O and δ13C. Whereas the period from c. 2600 cal bc to
2200 cal bc is relatively wet, the period from 2200 to 1800 cal bc is characterized by a strong rise in
especially δ18O, indicating a strong trend toward aridity. Between 1800 cal bc and 1500 cal bc, δ18O
and δ13C values remain relatively stable, and the variation remains within a range that is on average higher than during the preceding 1000 years. Thus, the Soreq Cave speleothems show that the
third and early second millennium bc climate is characterized by severe fluctuations, and eventually
settled into a drier and more stable climate around 2000 cal bc.
17
2.3.2 Lake Van
The sediments of Lake Van, southeast Turkey, have provided important insights into Near Eastern
palaeoclimate. Their main importance in their having been dated by the radiocarbon-independent
method of varve counting. The Lake Van region is at the intersection of three different climatic regimes and should therefore be sensitive to climate changes in each of these regimes. Lake Van is a
terminal, soda-rich lake with poor animal and plant life. Sediment cores were obtained and studied
in the 1970s (van Zeist and Woldring 1978) and 1990s (Landmann et al. 1996; Lemcke and Sturm
1997; Wick et al. 2003). The 1990s investigations have considerably revised the 1970s palaeoclimate reconstructions. The varves at Lake Van consist of annual carbonate and organic layers that
can be almost continuously counted for the last 14000 years. The sediment cores from Lake Van
provided a number of different palaeoclimate records, including pollen, charcoal, CaCO 3, δ18O,
and Mg/Ca records corresponding to various aspects of the past environment.
The Mg/Ca record has been shown to indicate changes in the salinity of the lake water with
respect to the current Mg/Ca ratio (Lemcke and Sturm 1997: 669). A high Mg/Ca ratio indicates
high salinity, whereas a low ratio indicates low salinity. The δ18O value depends on temperature,
humidity, and water influx into the lake. In order to detect the correlation between these variables
and δ18O, a model was developed based on data from the last 60 years. The modelled δ 18O values
were compared to the δ18O values recorded for that period, and a good correlation was found. As
in Soreq Cave, a high δ18O ratio corresponds to increased aridity. The vegetation development as
reconstructed from the 1990s sediment cores corresponds to the reconstruction by van Zeist and
Woldring (1978), but shows that their chronology should be pushed backwards by several thousand years. Pollen Assemblage Zones (PAZ) V-1 to V-9 broadly indicate the vegetation responses
to climate change and human impact in the region.
Together, these records indicate the following climate trends. A late glacial dry and arid climate
was replaced by more humid conditions at the onset of the Holocene. At c. 6000 cal bc, relative
humidity at Lake Van increased, in contrast to the globally documented trend toward a drier climate. Between 4000 and 2000 cal bc, relatively optimal climatic conditions persisted in the Lake
Van region, with Mg/Ca and δ18O values reaching minima around 2500 cal bc (Fig. 2.7B. The expansion of oak and minimal presence of Chenopodiaceae indicate forest-steppe vegetation (PAZ
V-7). The δ18O ratio between c. 2500 and c. 2200 cal bc indicates a drier climate, followed by a brief
reversal around 2100 cal bc. After 2100 cal bc, both the δ18O and the Mg/Ca ratios increase, pointing to a more arid climate, lasting for several centuries. At the same time, the importance of oak
diminished in favour of a more open landscape (PAZ V-8). Wick et al. argue that it is unlikely that
this reduction resulted from human activity, and was probably caused by summer drought. Slightly
later, around 1800 cal bc, human activity in the area became apparent through evidence for disturbance by livestock. Around 0 cal bc, modern climatic conditions were established, in tandem with
a further decline of oak (Lemcke and Sturm 1997; Wick et al. 2003).
2.3.3 Dead Sea
Several Near Eastern palaeoclimate records measure variations in the water level of the Dead Sea
(Frumkin et al. 2001; Enzel et al. 2003; Migowski et al. 2006). Most proxies measure variations in
the water level or in salt deposition of the Dead Sea. These variables are directly correlated with
the amount of water the Dead Sea receives, as well as local evapotranspiration, which in turn are
correlated with rainfall, humidity, and temperature. As the Dead Sea itself receives relatively little
rainfall, lake level changes directly reflect precipitation in the wetter parts of its drainage basin, especially the northern part. It has been demonstrated that Dead Sea water level changes are in step
with precipitation changes in and around the Jordan Valley. Precipitation is largely determined by
cyclones coming from the Mediterranean, and for dry years it can be shown that cyclones affect
areas to the north and west of the Levant, rather than the Levant itself. If this correlation holds
for the past as well, changes in storm tracks may explain some of the variety observed in the Dead
Sea water level (Enzel et al. 2003).
18
Reconstructed Dead Sea water level changes are shown in Fig. 2.7C. General trends are as follows, the caveat being that several temporal gaps in the reconstruction remain. The Younger Dryas
and the 6000 cal bc dry event are clearly detected by the long sequence recovered by Migowski et
al. (2006). Wet phases are reconstructed for 8000–6600 cal bc and 5600–3500 cal bc. These periods are, however, interspersed with very abrupt and short-lived reversals in climate. Major dry
events occurred at 8200, 6600, 2200, and 1500 cal bc. The drop around 2200 cal bc is corroborated
by Enzel et al. (2003), although they do not detect the return to wet conditions reconstructed by
Migowski et al. Finally, Frumkin et al. (2001) also reconstruct several wet phases, dated to 83006000 cal bc, 4000–2900 cal bc, and 1400–800 cal bc. These phases, again, only partly overlap with
the data presented above, but they do suggest that the late third–early second millennium experienced a relatively dry climate, resulting in a Dead Sea water level lower than 400 m below sealevel, compared to -390 m before the introduction of modern large-scale mechanized irrigation
schemes.
2.3.4 Gulf of Oman and Red Sea
Winds transport significant amounts of dust from Mesopotamia and the Zagros toward the
Persian Gulf and the Gulf of Oman. The mineral composition of this dust has been analysed
to determine the provenience of this dust, resulting in the recognition of a mineral signature of
Mesopotamian dust. Hence, sediment cores from the Persian Gulf and the Gulf of Oman provide
information on the amount of Mesopotamian dust present. Changes in dust deposition correlate
directly with past aridity changes in Mesopotamia, even though climate in the Gulf of Oman itself is not governed by the same mechanisms as Mesopotamia itself. To investigate these changes,
Cullen et al. (2000) sampled a sediment core from the Gulf of Oman at regular intervals. Analyzed
sample variables included oxygen isotopes, calcium carbonate and dolomite content, and strontium (Sr) and neodymium (Nd) isotope composition. Calcium carbonate and dolomite indicate
the amount of dust in the core. The Sr/Nd isotope composition analysis confirmed that most of
the dust deposited here derived from Mesopotamia. Several accelerator mass spectrometry (AMS)
radiocarbon dates provided an age control for the reconstructed sedimentation rate. The Gulf of
Oman core documents dust deposition for the last 25000 years. The Younger Dryas cold period
is evidenced by a high content of wind-borne dust. A second dust peak is recorded for c. 3200 cal
bc. Finally, a major dust peak appears at c. 2100 cal bc, levelling off until c. 1700 cal bc, dated by
four AMS 14C-dates (Fig. 2.7D).
A similar climatic anomaly is recorded in a core from the Shaban Deep in the northern Red Sea,
which is located between the rainfall regimes of the Mediterranean and the Afro-Asian monsoon.
The Shaban Deep is a basin where brine is deposited annually. The laminated brine sediments provide a record for the climatic conditions under which deposition occurred. Brine sedimentation is
influenced by climate related surface water changes. The sediments have been cored, and the upper
65 cm of the undisturbed sediments document the sedimentation sequence from 4000 to 1900 cal
bc. Chronological control comes courtesy of six AMS radiocarbon dates. Various proxies, including δ18O values, are analysed; all detecting a climatic anomaly between 2200 and 2000 cal bc (Fig.
2.7E). The δ18O values show a distinct peak between 2200 and 2000 cal bc, indicating unfavourable
climatic conditions, primarily interpreted as an arid event (Arz et al. 2006).
2.3.5 Palaeoclimate records from northern Mesopotamia
Geoarchaeological investigations at ancient river beds, pits dug for irrigation canals, and test pits
around Kazane Höyük, Turkey, document past environmental change (Rosen 1997). The presence
of datable artefacts in sediments afforded some crude chronological control (Fig. 2.7F). During
the Late Chalcolithic and early EB periods, relatively favourable environmental conditions prevailed at Kazane Höyük. The region was more heavily forested than at present, river discharge
and rainfall were more regular, and the groundwater table was higher. This favourable period was
followed by a dry phase probably starting slightly before the end of the third millennium bc and
continuing well into the second millennium bc. This phase was characterized by ‘accelerated chan-
19
nel erosion and desiccated abandoned floodplains’ (Rosen 1997: 405). Rosen suggests that climate
change, deforestation, or diversion of local rivers for irrigation purposes could explain this erosion phase. Climate change is put forward as the most likely cause, given the fact that there was
no return to previous environmental conditions when settlement in the area, and hence human
action, was resumed. There is thus evidence for late third–early second millennium bc desiccation
accompanied by seasonally bound rainfall at Kazane Höyük.
The Pre-Pottery Neolithic site of Göbekli Tepe, Turkey, was recently subjected to micromorphological investigations to reconstruct local climate conditions. Pustovoytov et al. (2007) investigated pedogenic carbonate coatings on stone walls, resulting in δ18O and δ13C curves for the period from 8000 to 2000 cal bc. that correspond to air temperature and the ratio of C3/C4 plants,
respectively. During the early Holocene (8000–4000 cal bc), there was a decrease in δ13C and an
increase in δ18O values, and both curves showed opposite peaks that were most pronounced during
the earlier part of the covered time period. These curves suggest that during the period 8000–4000
cal bc, the climate was relatively humid, and that there was a development toward a more stable
climate with less pronounced oscillations. For the period 4000–2000 cal bc, the carbonate coatings indicate relatively high humidity and high temperatures, with a short-lived peak slightly before
2000 cal bc toward either higher humidity, or lower temperatures. After 2000 cal bc, there was a
trend towards lower humidity than recorded for the early and mid-Holocene periods (Fig. 2.7G).
Several geomorphological studies documented past climate changes in the Upper Khabur basin. Courty (1994) identified six alluviation and erosion phases in a geomorphological study of
the Wadi Avedji in the Upper Khabur. These phases corresponded to different climatic conditions
during the Holocene (Fig. 2.7H). During the early Holocene phase 1, there was evidence for strong
alluviation. Phase 1 was characterized by a climate warmer than that of today, followed by phase
2 when a trend toward higher aridity set in. From 5800 to 3800 cal bc (phase 3), climate improved
and humidity was high, followed by phases 4–5 with clear evidence for environmental change.
During phase 4 (3800–2250 cal bc), the Wadi Avedji incised into the floodplain created during the
previous phase, and water flow was irregular. Phase 4 (3800–2250 cal bc) also yielded evidence for
human river management. Phase 5 (2250–1850 cal bc) was characterized by even more severe climatic deterioration than in the previous phase. Finally, during Phase 6 there was a restoration of
climate, and present climatic conditions were established.
Further evidence for late third millennium aridity comes from recent geomorphological research by Deckers and Riehl (2007). They investigated fluvial deposits of the Wadi Jaghjagh and
Wadi Khanzir in the Upper Khabur basin (Fig. 2.7I). Botanical samples from these deposits were
used to support the palaeoenvironmental reconstruction. The presence of pottery sherds in these
deposits allowed chronological control through thermoluminescence (TL) dating. The resulting
reconstruction of the past environment stretched back to c. 5000 cal bc. During the late fourth
and early third millennium bc, the Wadi Jaghjagh seems to have experienced a relatively steady flow.
Botanical samples from the Wadi Jaghjagh section indicated that a riverine gallery forest with Salix
existed during this period. This material furthermore indicated the presence of woodland vegetation. After 2500 cal bc, these favourable environmental conditions ended, and discharge of the
Wadi Jaghjagh dropped. The authors suggested water extraction for irrigation purposes, increased
erosion resulting from the removal of the woodland cover, or late third millennium aridity as possible causes.
Finally, a similar, though undated shift was observed for the Wadi Jarrah in the Upper Khabur
basin. There, a channel system characterized by meandering channels was replaced by a braided
system, which was in turn replaced by a meandering system. As meandering channel systems result
from more regular flow regimes, the shift to a braided system possibly indicated drier conditions
in which rainfall was restricted to fewer, but heavier rain showers (Besonen and Cremaschi n.d.).
In the face of the evidence presented above on sequences in nearby wadi systems, it seems reasonable to assume that the change to a drier environment in the Wadi Jarrah also dated to the second
half of the third millennium bc.
The Leilan Climate Change Model (LCCM) was based on archaeological and geomorphological
evidence from Tell Leilan and Abu Hgeira (Weiss et al. 1993). The LCCM consisted of a four-stage
model of state development and environmental change (Fig. 2.7J, Table 2.3). Stage 1 of this model
20
corresponded to Leilan IIId (2600–2400 cal bc). During this period, the previously stable and relatively wet local alluvial regime was replaced by a more erratic regime. Water discharge and precipitation became more seasonal. For the second stage (Leilan IIa, 2400–2300 cal bc), no climatic data
were presented. For stage 3 (Leilan IIb, 2300–2200 cal bc), geomorphological samples indicated
continuation of the drying trend observed for stage 1, and a temperature that was slightly higher
than that of today. Stage 4 (2200–1900 cal bc) was called ‘Khabur hiatus 1’, based on the apparent
absence of archaeological sites for this period. Khabur hiatus 1 thus coincided with the climatic
deterioration of Courty’s phase 5. Based on geomorphological analysis of stratigraphic layers at
Tell Leilan, Weiss et al. suggested a three-phase model for the abrupt climate change of Khabur hiatus 1. Phase 1 was characterized by two distinct layers. The lower and oldest layer was 0.5 cm thick
and consisted of ‘weakly altered fine silt-sized volcanic glass with a few potassic feldspars, phytoliths, and rounded calcitic sand-sized pellets’ (Weiss et al. 1993: 1000). On top of this tephra layer
were Aeolian sediments mixed with volcanic ash and mudbrick, suggested to represent an ‘abrupt
arid phase marked by intensification of wind circulation’ (Weiss et al. 1993: 1001) and a higher
amount of dust in the air (phase 2). Phase 3, finally, was characterized by ‘the augmentation of soil
moisture, a reestablishment of marked dry-wet seasonal contrast, a lessening of heavy rainstorms
and a progressive soil stabilization’ (Weiss et al. 1993: 1000). A volcanic eruption was invoked as a
cause of the tephra deposition and the prolonged dust veil. Other palaeoclimate proxy records did
not, however, detect a major volcanic eruption around 2200 cal bc. It was therefore suggested that
the tephra at Tell Leilan resulted from a small-scale volcanic eruption in the ‘Anatolian–Caucasian
area’ that would not show up in far-off proxy records such as the GISP2 ice cores.
In a recent geomorphological study of sediments from Tell Brak, Courty (2001) redefined
the palaeoclimate sequence of the LCCM. According to this study, specific soil particles within a
unique stratigraphic sequence recorded at Tell Brak acted as the fingerprint of a major ‘dust fallout event’. This event was argued to be the same as, and therefore contemporaneous with, LCCM
stage 4. This claim of contemporaneity was based on the similarity of the reconstructed event scenarios. At both Tell Brak and Tell Leilan, an event was reconstructed involving large amounts of
dust being ejected into the air, followed by a period of major wind circulation. Analysis of the soil
particles at Tell Brak, however, indicated that this event could not have been caused by volcanism,
as originally argued. Courty considered as a working hypothesis the possibility that the dust fallout
resulted from a ‘violent air blast that caused multi-site ignition and widespread wild-fires’ (2001:
368). Furthermore, and more importantly, it was suggested that the event should be assigned a new
date. Based on the stratigraphy of Tell Brak, Courty argued that the event should not be dated to
2200 cal bc, but some 150 years earlier, to 2350 cal bc.
Several comments can be made about the palaeoenvironmental part of the LCCM, and Courty’s
adjustments. First, it should be noted that the sediments indicating the dust fallout event were never observed outside cultural contexts, even though wadi sections dating to this period have been
investigated in the Upper Khabur area. Second, it is not clear whether the 2200–1900 cal bc tephra
and Aeolian sediments were only occasionally observed in Tell Leilan trenches, or in every trench
reaching the appropriate levels of the Akkadian/post-Akkadian/Ur III periods. Third, some excavations with Akkadian/post-Akkadian/Ur III or contemporary levels do not yield any evidence
stage
date bc
Leilan
social/political
climate
4: collapse
2200–1900
IIc
Leilan abandoned, regional abandonments, Khabur hiatus 1
abrupt climate change (phase 1: tephra;
phase 2: marked aridity; phase 3: climate
stabilization)
3: imperialization
2300–2200
IIb
Akkadian control of Khabur (Brak),
water management
climate similar to the present; high
seasonal contrasts; higher temperature?
2: consolidation
2400–2300
IIa
Leilan acropolis city wall, pottery
mass-production
?
1: secondary state formation
2600–2400
IIId
Leilan 15>100 ha, centralization,
administration
increased seasonality, higher evaporation,
reduced discharge
Table 2.3: Summary of the Leilan Climate Change Model (LCCM) as described in Weiss et al. 1993.
21
for tephra layers, suggesting an extremely local rather than regional or even global cause and scale
for the dust fallout sediments observed at Tell Leilan (e.g. Meijer 2007: n. 16, on Tell Hammam etTurkman).11 Fourth, the sequence of environmental change observed at Tell Leilan and Tell Brak
is difficult to attribute to a single event of short duration, be it a volcanic eruption or a cosmic impact. In fact, Weiss et al. pointed to several inconsistencies between the LCCM and what might be
expected for volcanic eruptions, and already suggested that ultimately some other climatic mechanism may have been responsible for the observed scenario of environmental change (1993: 1002).
Finally, in a micromorphological study on sediments from Tell Brak, Matthews (2003: 385) argued
that the soil particles supposedly signifying the tephra deposition event are common in sediments
from multiple periods, and are thus unlikely to point to any single catastrophic event.
It might therefore be suggested that there is evidence for two distinct processes at Tell Leilan.
First, there is evidence for an as yet unidentified process that led to the deposition of a thin soil
layer of supposedly volcanic origin. Matthews’ study on the sediments from Tell Brak has already
shown that this origin may be incorrect. Furthermore, an eruption is an event of very short duration, with effects that last a few years at most (Stothers 1999). Such an event is therefore unlikely
to cause a 300-year long drought or dust-veil. Second, there is evidence for prolonged wind activity
and large amounts of wind-borne dust, pointing to abandonment of the site and perhaps drought.
As indicated by Weiss et al., it is difficult to find a single explanation for these observations. It
therefore seems most logical to disentangle both processes. In other words, the event leading to
tephra deposition and the subsequent 300-year drought are unrelated (cf. Courty and Weiss 1997:
143). Summarizing, considerable doubt as to the date and nature of the dust fallout event documented at Tell Brak and Tell Leilan remains. On the other hand, the evidence for a long-term process toward higher aridity, Courty’s stages 4–5, is stronger, especially because it derives from both
cultural and natural deposits. This trend is commensurate with the evidence for increased aridity
at Tell Leilan, if regarded separately from the tephra evidence.
2.3.6 Reconstructing climate change
It should be noted that the interpretation of palaeoclimate data and their extrapolation to other
areas of the Near East remain problematic for three reasons. Firstly, data come from a variety of
sources and geographically widely separated locations (Table 2.2, Figs. 2.6, 2.7). This scattering of
data types and locations sometimes results in apparently contradictory palaeoclimate reconstructions. Secondly, some proxy records will fail to record short-term climate changes due to their low
resolution. Such short-term climate change events may also remain undetected in high-resolution
proxy records elsewhere due to the sometimes limited geographical distribution of these events.
Finally, palaeoclimate reconstructions are complicated by the fact that climate-forcing mechanisms
may have been different in the past, so that modern data cannot be directly extrapolated to the
past. Thus, palaeoclimate reconstructions for the Near East can presently only achieve a limited
precision. Gaps in local climate histories are now being filled through the modelling of palaeoclimate developments (e.g. Bryson and Bryson 1997; Riehl et al. 2008).
The geographical locations of Near Eastern climate proxy records highly influence the resolution that can be achieved for palaeoclimate reconstructions. For some areas, multiple records
are available that can corroborate each others’ respective results. In other regions, climate proxy
records are scarce, and palaeoclimate can only be inferred from records elsewhere. The geographical distribution of climate proxy records is determined by the specific conditions under which
such records are created and survive. From Table 2.2 and Fig. 2.7 it is evident that well-dated,
high-resolution climate proxy records from northern Mesopotamia are lacking. This means that
any reconstruction can only be an approximation, and that some details must be inferred from
current climatic conditions.
11 It should be noted, though, that this absence could at least partly be explained by the possibility that excavators have
not been actively looking for this layer, or that the geomorphological analyses that are necessary to identify it have not
been carried out.
22
1200
1100
1000
900
800
mm
700
600
500
400
300
200
2004/2005
2003/2004
2002/2003
2001/2002
2000/2001
1999/2000
1998/1999
1997/1998
1996/1997
1995/1996
1994/1995
1993/1994
1992/1993
1991/1992
1990/1991
1989/1990
1988/1989
0
1986/1987
1987/1988
100
Fig. 2.8: Annual rainfall in hydrological years for the Jezirah (dark shading, average from Hasakeh, Qamishli, and Raqqa, cf. Fig.
2.4) and Jerusalem from 1986 to 2005 (light shading, Jerusalem data from the Israel Meteorological Service).
Currently, the best climate data for the Near East come from Soreq Cave and Lake Van.
Climate in these locations is generally driven by the same mechanisms. Rainfall in all locations is
mainly deposited during winter by cyclones from the Mediterranean, as is the case for northern
Mesopotamia. For Soreq Cave, it has been shown that changes in humidity and sea-level respectively correlate well with precipitation changes in Jerusalem. Modelling of annual precipitation for
the last 7000 years in both Jerusalem and Qamishli indicated similar trends on a very general level,
although details and timing may be considerably off (Bryson and Bryson 1997: 579). A comparison
of recent precipitation data seems to suggest that there are similarities and that dry years generally occur simultaneously. However, given the weak correlation between precipitation in Jerusalem
and in the Jezirah (rs=0.48), it is clear that data from Jerusalem must be cautiously extrapolated to
northern Mesopotamia (Fig. 2.8).
The palaeoclimate proxy records discussed above document two types of changes. First, a
number of records, especially those from higher latitudes, record the gradual replacement of relatively favourable third millennium bc climatic conditions by a more arid climate with either lower
or more seasonally confined rainfall. Second, several records document an abrupt arid event lasting
several hundred years at most and followed by a return to pre-event climatic conditions. Examples
of the first include the Soreq Cave, Lake Van, and Harran Plain records, as well as the geomorphological evidence from the Upper Khabur Basin, except for the evidence from Tell Leilan. An
abrupt arid event is recorded in the Gulf of Oman and Red Sea records, and less clearly in the
Dead Sea record. Several researchers have noted this discrepancy between an abrupt arid event and
a long trend toward higher aridity, but this discrepancy has not yet been fully explained (e.g. Arz et
al. 2006: 438–40; Magny et al. 2009: 831).
Interestingly, some palynological studies in the Near East fail to detect the late third millennium climate change (e.g. Gremmen and Bottema 1991; Bottema 1997; Neumann et al. 2006). Both
Gremmen and Bottema (1991) and Neumann et al. (2006) argued that climate remained largely
unchanged for the last 6000 years, and that human impact on the environment must therefore be
considered more significant. The pollen core from Lake Bouara showed some minor evidence for a
dry episode around 2000 bc, but the evidence from pollen cores from Anatolia and the Zagros area
was less conclusive (Bottema 1997: 509–10, 514). The vegetation reconstruction from the Golan
(Neumann et al. 2006) indicated a high amount of arboreal pollen for the Early/Middle Bronze
Age, indicative of Mediterranean climatic conditions. However, there are many variables involved
in the response of vegetation to climate change, so that deducing regional climate change from
localized pollen spectra remains problematic. Plant species react differently to climate change, with
short-lived species showing the strongest reactions. Furthermore, the number of pollen spectra
23
from northern Mesopotamia is limited. Finally, dating is relatively coarse and usually derived from
a few radiocarbon dates, from which a sedimentation rate is extrapolated. These dates may be biased as a result of carbon-depletion in water-logged sites.
2.3.7 Synthesis of climate change
The combined evidence from the palaeoclimate proxy records discussed here suggests the following scenario of climate change. The fourth millennium is subject to considerable fluctuations. The
Soreq Cave speleothems document notable shifts in stable oxygen isotope levels for the period
3500–3000 cal bc. This unstable climate is also reflected in the Dead Sea record. Notable drops in
water-level are recorded for c. 3700 cal bc and around 3500 cal bc, coinciding with a major δ18O
peak in Soreq Cave. The Gulf of Oman core shows a minor rise in dolomite and CaCO3 percentages slightly before 3000 cal bc, indicating a higher dust-content, which again is paralleled in Soreq
Cave. Wadi flow regimes in the Jezirah are regular, indicating that rainfall is more evenly spread
over the year than is the case today. Also, low erosion rates imply a dense vegetation cover.
Climatic conditions in the Near East became more favourable after c. 3000 cal bc. For the period 3000–2500 cal bc, Soreq Cave and the Dead Sea document relatively wet conditions, although
at Soreq Cave short arid spikes occurred around 3000 and 2700 bc. The percentage of oak pollen in Lake Zeribar was still high, despite declining continuously from the fourth millennium onward (Stevens et al. 2001). Low pollen preservation inhibits any interpretation for this period at
Lake Mirabad, but around 2500 cal bc, the percentage of oak pollen was very high (Stevens et al.
2006).
Climate records across the Near East start to attest drier conditions from 2500 cal bc, with progressively more records indicating aridity towards the end of the third millennium. The Soreq Cave
speleothems signal gradually increasing δ18O for the period 2500–2000 cal bc, with δ13C following
from c. 2200 cal bc. At Lake Van, pollen evidence indicates drier climatic conditions towards the
end of PAZ 7. The δ18O isotope level around 2250 cal bc is the highest recorded for the third millennium bc. Lake Van shows a short return to higher humidity levels at c. 2150 cal bc, followed by
a gradual and stronger rise in both δ18O and Mg/Ca. The records most prominently indicating a
sudden arid event, that is the Gulf of Oman and Red Sea cores, are dated between 2200 and 2000
cal bc. Several records indicate a short-lived return to wetter conditions around 2000 cal bc, followed by more prolonged aridification. After 2000 cal bc, some records show a return to previous
wetter conditions, including the Dead Sea and the Gulf of Oman. Most records indicate, however, that the late third millennium climate change is of a more permanent nature, signifying the
onset of more or less modern climatic conditions. This seems for example to be the case for the
Wadi Jaghjagh, Göbekli Tepe, and Kazane Höyük records. For the second millennium, virtually all
records indicate drier conditions than existed previously during the fourth and early third millennium (Kuzucuoğlu 2007a: 474). Whether the combined signals should be understood as indicating
one prolonged arid phase, or two smaller arid events punctuated by a short humid period as argued
by Kuzucuoğlu (2007a: 476) remains unclear on the basis of the available northern Mesopotamian
evidence. Modelling of northern Mesopotamian climate during this period seems to confirm that
there was one prolonged dry period, although there are clear local differences (Riehl et al. 2008:
fig. 2).
Summarizing, four stages of different climatic conditions can be established. The first stage
lasted from 4000–3000 cal bc and was characterized by considerable climatic fluctuations. Notable
dry spells occurred around 3700, 3500, 3300, and 3000 cal bc. The second stage started c. 3000 cal
bc and was relatively wet across the Near East. Around 2500 cal bc, many proxy records start to
indicate drier climatic conditions, initiating stage 3 (2500–2000 cal bc). This stage was characterized by progressively drier conditions. A possible short humid period may have occurred around
2000 cal bc, but this period has not been detected in all proxy records. Stages 1–3 fall into Rosen’s
Middle Holocene phase of highly erratic precipitation changes (2007: fig. 3.5). Stage 4, from 2000
cal bc onwards, was characterized by climatic signals that vary across the Near East. Some records
indicate a return to wetter climatic conditions as persisted during the first half of the third millennium, whereas other records suggest that the climate remained dry, similar to the conditions
24
existing today. The Dead Sea and Gulf of Oman records suggest a more or less rapid return to
previous or even wetter conditions. At least for the Dead Sea this return is relatively short-lived, as
the water level drops significantly around 1500 cal bc. The evidence from Lake Van, Soreq Cave,
the Harran Plain, and the Wadi Jaghjagh on the other hand indicates that the late third millennium
arid phase continues well into the second millennium.
2.3.8 Evidence for drought
The foregoing overview of northern Mesopotamian palaeoclimate has highlighted the occurrence
of an arid phase at the end of the third millennium and the beginning of the second millennium
bc. The establishment of an arid climate is, however, not in itself evidence for drought, which is
a shortage of water for the inhabitants of a given region. It may well be that overall precipitation,
humidity, and river discharges declined, but that there still was sufficient water for both humans
and their crops and livestock. Although changes in the composition of the crop spectrum are regularly linked to local environmental changes, it should be noted that such changes may also come
about because of changing political, socio-economic, or cultural conditions (Riehl 2009: 18–19).
A change in the crop spectrum can for example result from changing food habits rather than be
down to any particular environmental reason. Thus, the development of the crop spectrum from
the third to the second millennium as increasingly reflecting arid conditions is likewise not in itself
evidence of drought. An independent method is therefore required to test whether or not particular changes in crop selection were correlated to the occurrence of water stress or drought.
In order to address this issue, the stable carbon isotope composition has been measured from
grain samples collected from numerous Early and Middle Bronze Age sites across the Levant and
northern Mesopotamia (Riehl 2008; Riehl et al. 2008).12 It has been observed that a correlation
exists between the stable carbon isotope composition of C3 plants and water availability through
irrigation, soil moisture, or humidity. C3 plants have lower carbon fixation in conditions of low
water availability, resulting in an increase in δ13C or a decrease in Δ13C. Even though there are other
variables influencing Δ13C, water availability seems to be the most important environmental variable (Riehl et al. 2008: 1012). Because Δ13C values can only be interpreted when data are available
on the local moisture conditions, Riehl et al. used a macrophysical climate model (MCM) based
on data from Lake Van and Soreq Cave to simulate ancient climate conditions. The resulting local
climate conditions were compared to the stable carbon isotope compositions of barley under the
assumption that this crop was generally cultivated without additional irrigation. Comparison indicated that different Δ13C values were recorded for sites in different geographical settings and with
different moisture conditions, that the Δ13C values of other cereal crops showed similar trends to
those of barley, and that a general decrease of Δ13C values coincided with reconstructed increasing
aridity at the end of the third millennium, and especially during the Middle Bronze Age (Fig. 2.7K)
(Riehl et al. 2008: 1020). These observations, then, suggest that there was indeed a certain level
of water stress or drought at various sites in northern Mesopotamia, although the severity of this
water stress differed per site. This conclusion bolsters claims that crop selection was altered due
to environmental change, and legitimizes the investigation of other responses to environmental
change in general and drought in particular.
A similar study has recently been carried out on subfossil wood from a Tamarix tree from
Mount Sedom, on the southern shore of the Dead Sea (Frumkin 2009). In this arid region, temperature and precipitation are the main factors constraining plant growth, and this is reflected in
the 13C and 15N isotope composition of plants. Whereas 13C enrichment can clearly be correlated
to high temperatures and/or low precipitation, 15N enrichment is primarily associated with low
precipitation, although problems with this correlation remain. Using the correlation between 13C
isotopes and rainfall as it has been observed in modern Tamarix trees in the region, the absolute
precipitation levels for the period encompassed by the ancient tree trunk could be reconstructed.
Radiocarbon dates established a growth period between c. 2300 and 1930 cal bc, at which time
12 These sites were Hirbet ez-Zeraqoun (Jordan), Tell Fadous (Lebanon), Qatna, Alalakh, Tell el ‘Abd, Emar, and Tell
Mozan (Syria).
25
the tree died. During this period, the 13C and 15N isotope levels in the sampled tree show similar
trends, which can be summarized as follows (Fig. 2.7L). Precipitation was highest at the start of
this palaeoclimate record, and remained so with some minor fluctuations until 2200 cal bc. At that
time, there occurred a drop in precipitation lasting some 70 years, after which rainfall briefly returned to the previous level. After 2130 cal bc, precipitation gradually dropped to a level similar
to or even slightly lower than that of today, with an additional short-lived arid event taking place
at 2020 bc. The analysed tree died around 1930 cal bc. This is in agreement with the current situation, in which almost no Tamarix trees survive in the region. This palaeoclimate record therefore
strongly suggests that plants were significantly affected by aridification during the late third and
early second millennium bc.
Recently, evidence was presented for subsistence stress among humans during the late third to
early second millennium bc. At Tell Barri, a number of human skeletons have been studied from
EB II/III to Neo-Assyrian levels, including a sample that could be specifically dated to transitional
levels between the Early and Middle Bronze Ages. Although the total assemblage consisted of
only 92 individuals, clear differences between each period appeared. The study of teeth allowed
the reconstruction of the frequency of enamel hypoplasia and dental caries. Enamel hypoplasia is
linked to nutritional stress during childhood, whereas the frequency of dental caries is an indicator of food quality. Skeletons from the EB IV/MB I and the Neo-Assyrian sub-samples showed
the highest frequencies of enamel hypoplasia, pointing to a lower nutritional status during these
periods. Similarly, dental caries was especially high in the EB II–III and EB IV/MB I sub-samples,
again indicating subsistence stress during these periods. Finally, these sub-samples also showed a
high level of child mortality that seemed to deviate statistically from the expected level (Sołtysiak
forthcoming). Thus, the combined mortuary evidence suggested that the late Early Bronze Age
period was marked by higher than normal infant mortality, combined with nutritional stress that
was particularly evident during the EB IV/MB I transitional period (Fig. 2.7M). Despite the limited nature of the evidence, that is, the small sample and the fact that it came from a single site, it
seems to accord relatively well with the indications for subsistence stress that have emerged from
the palaeobotanical and isotopic evidence.
2.4 Chronology construction
Correlation of environmental developments with material culture change relies on precise absolute dating of both strands of evidence. Dating environmental developments almost invariably
involves radiocarbon dating and/or similar methods from the natural sciences. Dating material
culture change, on the other hand, involves a large array of dating methods including radiocarbon
dating, historical dates, and typochronology. This difference in dating methods presents archaeologists with considerable difficulties (Bruins 2001: 1148). Correlation of these diverse methods
has proved extremely complex, leading Reade (2001: 1) to observe that ‘most scholars are averse
to chronological studies’. This aversion often led to the ready acceptance of a convenient, wellestablished chronology, in this case the so-called Middle Chronology that dates the fall of Babylon
by the Hittite king Muršili I to 1595 bc, even though this chronology is flawed in several respects. 13
There are now several studies that critically address established chronological conventions. These
studies focus on the implications of the increasingly refined correlation between text passages
describing astronomical observations and historical events and reigns (Gasche et al. 1998; Reade
2001).
At the basis of the Middle Chronology are Brinkman’s chronological charts (1977). This chronology is based on a number of synchronisms, several copies of king lists, astronomical observations, and multiple assumptions as to calendars that were used, the methods by which regnal years
were counted, and the lengths of individual reigns and periods. For example, the starting point of
the reign of Sargon of Akkad largely depends on the length that is assigned to the so-called Gutian
13 Pruß for example observed that he ‘has never met a colleague dealing with chronology, who actually believes in the
correctness of the Middle Chronology’ (2004, n. 94). Zettler wrote in a similar vein of the Middle Chronology that it
‘has come to be cited more out of convenience than conviction by archaeologists and historians’ (2003: 20).
26
period, variously established at between 30 and 100 years, and on whether 37 or 56 regnal years
are assigned to Naram-Sin (Sallaberger 2004a).14 Furthermore, the king lists may have sequentially
ordered periods that actually overlapped or that were contemporary. The main difficulty in departing from the Middle Chronology is that it has widespread repercussions for almost all other areas
and sites in the wider Near East, as well as for many other periods. This has led many researchers
to the continued, albeit reluctant acceptance of the Middle Chronology (e.g. Lebeau 2000a: 170;
Akkermans and Schwartz 2003: 12).
The Middle Chronology is largely based on historical sources coming from outside the Jezirah.
Although there are texts from, as well as about, the Jezirah for the late third millennium, periodization for the Jezirah during this period is mainly achieved through the study of stylistic variation, stratigraphy, and radiocarbon dates. The correlation of historical events with radiocarbon
dates, stratigraphic evidence and material culture is notoriously difficult. The term Akkadian may
for example refer to the period during which a political entity of that name existed, or only to the
period for which Akkadian influence in the north is archaeologically detectable, as at Tell Brak or
Tell Mozan. Additionally, it has proven extremely difficult, if not outright impossible to define
material remains as representing a typically Akkadian culture. It has recently been shown that many
pottery types thought to be typically Akkadian actually had much longer life-spans, thereby further
weakening the link between political and cultural designations (Zettler 2003: 27–8; Nicolle 2006:
166). Recently, however, important historical synchronisms between Mari, Ebla, and Beydar have
been established, allowing a finer control of archaeological periods (Sallaberger 1998; Archi and
Biga 2003).
The development of the discipline of Near Eastern archaeology has presented further problems for periodization and synchronization in the Jezirah. For a long time, the northern Jezirah
in itself has not been a research area. Leaving aside Mallowan’s pioneering work in the 1930s, the
region has always been overlooked in favour of southern Mesopotamia and western Syria. This is
reflected in the preferential use of western Syrian chronologies in the western part of the Jezirah,
and traditional Mesopotamian divisions in the Khabur area. The artificial division that is thus created possibly reflects real variations between the western and eastern parts of the Jezirah, but may
just as well be imposing these variations upon otherwise similar material. This division is difficult
to break down (see Lebeau 2000a: 168–9).
Recent attempts at chronological refinement have taken these considerations into account,
resulting in the Early Jezirah (EJ) and Old Jezirah (OJ) periodizations that rely on stratified material culture change and radiocarbon dates (Pfälzner 1998; Lebeau 2000a; Dohmann-Pfälzner and
Pfälzner 2001). The EJ periodization, including its regional subdivisions, builds upon the work of
multiple excavations across the Jezirah. It aims to establish a framework within which the effects
of chronological and regional variation can be clearly discerned and separated (Lebeau 2000a: 170).
The OJ periodization, introduced at Tell Mozan for the early second millennium bc (DohmannPfälzner and Pfälzner 2001), does not seem to be as widely accepted as the Early Jezirah chronology yet. Although this new chronology has removed the ambiguous and often misleading southern
Mesopotamian terminology from the archaeological discourse in the Jezirah, it should be realized
that the new chronology still has many ties with the south. A new terminology masks these ties,
but does not eliminate them. It rather allows some stereotypical views to be corrected.
Rigorous analysis of multiple radiocarbon dates from secure stratigraphic contexts across
the Near East and the Mediterranean may ultimately lead to the revision of large parts of the
conventional chronological frameworks (Bruins 2001; Bruins et al. 2003a; Manning et al. 2006).
Unfortunately, third and second millennium bc Mesopotamia is only now being subjected to this
14 Radiocarbon dates from EJ III/IV contexts at Beydar, Brak, Mari, and Mozan are still too inconsistent to support
either a 40 or 100-year length for the Gutian period. For example, the 2175 cal bc date of phase 2a of the Tupkish
palace in Mozan fits only with a 40 year Gutian period using the Middle Chronology, as seems to be the case with
the 2291–2200 cal bc date for the destruction of Mari Ville II. On the other hand, the very early EJ III radiocarbon
dates from Tell Brak and Tell Beydar would favour a 100-year Gutian period, again following the Middle Chronology
(Sallaberger 2007: 421).
27
date BC Jezirah
1600
Middle
Jezirah I
Southern
Mesopotamia
Beydar
Brak
Chuera
1700
VIIc
Old
Babylonian
1800
1900
2000
2100
Old
Jezirah II-III
Old
Jezirah I
Early
Jezirah V
VIIb
lower town
administrative
complex
VIIa
P Old
Babylonian
Isin/
Larsa
Ur III
Gutian
IVb
2400
Early
Jezirah IV
Early
Jezirah IIIb
Akkadian
Early
Dynastic IIIb
IVa
III-IV
IIIb
N postAkkadian
2500
2600
2700
Early
Jezirah IIIa
Early
Jezirah II
Early
Dynastic IIIa
Early
Dynastic II
IIIa
VI west 3-4
M Akkadian
IE
L postNinevite 5
K late
Ninevite 5
II
ID
VI west 1-2
2900
3000
3100
0
North Jazira
Survey
VIIc
VIIb
Late MBA
I
VIIa
Khabur
Late EBA/
Early MBA
IIc
IIb
VIcd
IIa
IC
BirecikEuphrates
Dam Survey
VIIIa
city wall
VI east 5-8
Middle EBA
Late third
millennium
Early EBA
Ninevite 5
Northern
Uruk
VI east 3-4
IB
VI east 1-2
2800
Balikh
Survey
VI west 6
VI west 5
2200
2300
Leilan
VIIIa
VII/VIII
Q Mitanni
Kassite
Hammam
et-Turkman
IIId
IIIc
(I)
Early
Early
Jezirah I
Dynastic I
Early
Jezirah 0
Late Chalcolithic 5
Late Uruk
IIIb
IIIa
J early
Ninevite 5
VIab
H post-Uruk
G late Uruk
V
IV
V
Table 2.4: Comparison of third and early second millennium bc periodizations for the Jezirah, southern Mesopotamia, selected
sites, and the Balikh Survey, Birecik-Euphrates Dam Survey, and North Jazira Survey (based on Brinkman 1977; Curvers 1991;
Algaze et al. 1994; Wilkinson and Tucker 1995; Lebeau 2000a; Dohmann-Pfälzner and Pfälzner 2001; Oates et al. 2001; Pruß
2004; Ristvet 2005; Sallaberger 2007).
type of chronological reanalysis. The relatively subjective analysis of stylistic and textual linkages
therefore currently remains the most important method to assess the implications for Mesopotamia
of chronological revisions elsewhere.15
However, independent support for the Middle Chronology now comes from radiocarbon dated
wooden beams from Anatolia. The Aegean Dendrochronology Project (AEP) has constructed
and dated a 1500+ years floating tree-ring sequence from wooden beams recovered from various
archaeological sites in Anatolia. This sequence has been radiocarbon dated, allowing the proposal
of absolute dates for the structures from which the wood was taken, and from there the dating of
historical reigns and events associated with those structures. These structures include the Sarıkaya
Palace at Acemhöyük, and the Waršama Palace at Kültepe. For these buildings, construction dates
of 1774 +4/-7 bc, and 1832 +4/-7 bc are proposed, respectively. In both buildings sealings and
documents relating to Shamshi-Adad were found. Interpolation of these data with textual evidence
suggests that the Middle Chronology must be lowered by thirteen to seventeen years. According to
the authors, the High, Low, and Ultra-Low chronologies are now respectively rendered impossible,
unlikely, and very unlikely (Kuniholm et al. 1996: 782; Manning et al. 2001: 2534).16
15 For example, no collection of Mesopotamian radiocarbon dates featured in a recent special issue of Radiocarbon
devoted to Near Eastern chronology (see Bruins 2001: and references therein). The ARCANE project (Associated
Regional Chronologies for the Ancient Near East and the Eastern Mediterranean) that was initiated in 2002 aims
to bring together and synchronize regional culture/historical chronologies, as well as radiocarbon dates, in order
to achieve a reliable absolute chronology for the third millennium bc and the preceding and following centuries
(http://www.arcane.uni-tuebingen.de/index.html (Accessed 4 February, 2009)).
16 On the proposed bias towards too early dates for mid-Holocene radiocarbon determinations that would affect these
chronologies, see Keenan 2002, Manning et al. 2002, and Kuzucuoğlu 2007b.
28
In the present study, the Middle Chronology will be followed, although accumulating evidence
suggests that this chronology is in need of at least some minor revision. Nevertheless, this chronology remains the most widely employed, and thus better accommodates comparative study of
multiple regions and sites than newer chronologies that are generally only applied to single sites.
The debate on Middle versus Low Chronology still focuses on textual evidence, which makes extrapolation to archaeological periodizations difficult. The repercussions of the shorter historical
chronologies for material culture chronologies as yet remain to be determined. A step in the right
direction would be the full integration of historical data and stratified archaeological material, and
a stronger focus on radiocarbon dating and other dating methods from the natural sciences. Table
2.4 shows several chronologies that are currently in use in the Jezirah and neighbouring areas.
2.5 Some notes on the political history of northern Mesopotamia
Historical information on northern Mesopotamia is available from the twenty-fourth century bc
onwards in the form of the archives from Ebla palace G and Tell Beydar. During the c. 50 years
covered by these archives, the initial political dominance of Mari was challenged by a coalition of
Ebla, Kish, and Nagar (Tell Brak). Although the battle fought between these powers apparently
resulted in a victory for Ebla, the city was destroyed soon after. It has recently been suggested
that Mari was most likely the power responsible for this destruction, rather than Sargon of Akkad
(Archi and Biga 2003: 35). Considering that Ebla and Nagar were allies, the destruction of phase
L at Brak may probably also be attributed to Mari (Sallaberger 2007: 422). Again not much later,
Mari was destroyed, probably by Sargon of Akkad, who recorded that he subdued Mari and Elam
(Frayne 1993a: E2.1.1.1; Archi and Biga 2003: 34). The texts from Nabada (Tell Beydar) and Ebla
testify to the urban nature of settlement in the Upper Khabur region. Numerous cities are menShimanum?
Urshu?
Abarsal?
Khabura?
Urkesh
Shekhna
Carchemish
Taidu?
Nabada
Nagar
A zuhinum ?
Yamu diu m?
D i ba r
Emar
Tuttul
r
ša
Ba
Silver
Mountains Talmush
Ninua
Ebla
Cedar
Mountains
Akkad?
Kish E
lam
Mari
2000+ m
1500-2000 m
1000-1500 m
500-1000 m
0-500 m
0 km
intermittent lake
lake
N
30
60
Fig. 2.9: Map of northern Mesopotamia showing late third millennium bc names of sites and regions mentioned in this study.
29
tioned. Among the important, and probably independent, northern Mesopotamian centres of the
mid-third millennium were Nagar, Shekhna, Urkesh, Abarsal, and Azuhinum. These centres probably controlled the greater part of northern Mesopotamia (Ur 2004: 272) (Fig. 2.9). 17
Sargon of Akkad’s impact on the political situation of northern Mesopotamia is difficult to assess. He claimed to have reached Tuttul, the Cedar Mountain, and the Silver Mountains, that is Tell
Bi’a at the confluence of the Balikh with the Euphrates, the Amanus, and the Taurus Mountains,
respectively. However, it remains difficult to ascribe archaeologically observed destructions to his
actions, as the example of Ebla, mentioned above, clearly shows. Thus, links between Sargon and
archaeologically observed contemporary events or developments must be carefully considered
before they are drawn. In fact, it has not yet been settled whether Sargon’s western campaign was
a large-scale military operation with the purpose of conquering and subduing foreign lands, or
whether it was no more than a surprise military raid, taking advantage of already weakened opponents such as Mari and Ebla. The important observation must be that west of the Upper Khabur
region there are no archaeological indications of lasting Akkadian influence. 18
Possibly from the time of Sargon onwards but certainly at the time of Naram-Sin, the middle
and eastern Upper Khabur region must have been under Akkadian political control. The exact nature of this control remains uncertain, however. Tell Brak probably maintained a key position in
this region. Sargon’s successor Rimush was mentioned in inscriptions on two stone vessels found
at Tell Brak. It is difficult to see though how this relates to the position of Tell Brak at the time
of Rimush, because the vessels were found in the Naram-Sin Palace, and may therefore have been
brought to the site at a later time than the reign of Rimush. According to Sallaberger (2007: 424),
the vessels most likely represented gifts of Rimush to a local loyal governor and were therefore
heirlooms at the time of Naram-Sin.
Tell Brak was certainly an Akkadian stronghold during the reign of Naram-Sin. The NaramSin Palace, which should more properly be called a fortress, could be identified as such because
Naram-Sin’s name was stamped in mud bricks used for the construction (Frayne 1993a: E2.1.4.22).
The Akkadian influence can be inferred from an administrative text from the Naram-Sin Palace
listing cities and numbers of men, possibly soldiers (Eidem et al. 2001: 107). This list included,
among others, Urkesh, Shekhna, and Taidum. Another list of small amounts of animals, mostly
sheep, also named Shekhna (Eidem et al. 2001: 110). Urkesh is identified with Mozan (Buccellati
and Kelly-Buccellati 1999: 10), Shekhna with Tell Leilan (Charpin 1987), and Taidum possibly with
Tell Hamidiyeh (Eichler et al. 1985: 69; Wäfler 2001: 169). Except for Mozan/Urkesh, the relation
between the Akkadian administration at Brak and the cities in this list is unclear. The relatively
close proximity of Tell Hamidiyeh to Brak makes it likely that Hamidiyeh was under direct control
of Brak, yet not all cities in the list were, as is clear from Urkesh.
Despite the repeated claims that Shekhna/Leilan was an important ‘imperialized’ centre within
the Akkadian empire during Leilan period IIb (e.g. Weiss et al. 1993: 998), it is unclear what this
designation means exactly, and it must be considered a possibility that Leilan was of secondary importance with respect to Tell Brak, or that it was a vassal state like Urkesh. Much of the evidence
for Akkadian imperialization at and around Leilan is indirect, and the only direct evidence for a
form of Akkadian administrative presence at Leilan consisted of two sealings with Old Akkadian
inscriptions (Weiss 1997: 343; Weiss et al. 2002–3).19
At Mozan/Urkesh, sealings have been found that belonged to Tar’am Agade, a daughter of
Naram-Sin (Buccellati and Kelly-Buccellati 2000: 139). She was probably the wife of an unnamed
ruler or endan of Urkesh, known from another inscription found together with the sealing of
Tar’am Agade. It is likely that Urkesh was an ally of Akkad at this time and presumably also under Tupkish, a slightly earlier ruler of Urkesh. Thus, in the time of Naram-Sin, some parts of the
Upper Khabur region were controlled directly, as was probably the case at Tell Brak. In other areas,
17 See Sallaberger 2007 for a discussion of the problems associated with locating third-millennium toponyms. The location of Abarsal is especially problematic, and besides Kazane Höyük, a location along the Upper Euphrates has been
proposed as well (Bunnens 2007: 50).
18 In fact, even Mari seems to have played a relatively marginal role up to and during the reign of Naram-Sin (Sallaberger
2007: 430).
19 See Ur (2004: 73) for a more elaborate discussion of the evidence for Akkadian control at Tell Leilan.
30
Akkad’s influence may have been more subtle and indirect, through alliances and vassal states, as
reflected in the evidence from Tell Mozan. This reconstruction would support the description of
the Great Revolt against Naram-Sin, according to which the rulers of the Upper Land favoured
Naram-Sin and refused to support Amar-Girid (Sallaberger 2007: 427).
The situation to the west of the Upper Khabur remains unknown during Naram-Sin’s reign.
Naram-Sin claimed to have reached the Cedar Mountain and also mentioned Tuttul, but it is not
certain whether the western or eastern Tuttul is meant (Sallaberger 2007: 429, n. 69). The impact of Naram-Sin’s western campaign is as difficult to assess as the impact of Sargon’s earlier
campaign.
After Naram-Sin’s reign, historical evidence from southern Mesopotamia on northern
Mesopotamian political affairs practically ceased until the Ur III period. The royal names preserved at Nagar/Brak and Urkesh/Mozan, however, indicate that at least these two places were the
centres of local kingdoms. At Tell Brak, there is one post-Akkadian ruler, possibly contemporary
with Shar-kali-Sharri or slightly later, and at Mozan the line of rulers seems to have extended almost to the end of the third millennium bc, although the evidence is admittedly meagre. It should
be noted that the names of the rulers of both Nagar and Urkesh were Hurrian, possibly indicating the growing importance of this population group in northern Mesopotamia (Sallaberger 2007:
431-2) (see Section 8.3).
During the Ur III period (2112–2004 bc), northern Mesopotamia figured occasionally in historical texts. The Ur III state apparently devoted most of its attention and military power to the east,
as far as can be reconstructed from royal year names. Thus, military operations to the west as are
known for Sargon and Naram-Sin are unknown for any Ur III ruler. However, there were political
contacts with the areas to the north and northwest of Ur as well. There were marriages between
Ur III rulers and princesses from cities in the north, including Mari. The marriage with a princess
from Mari could be interpreted as an alliance with that city to create a political and military buffer
to the north of the Ur III state (van de Mieroop 2007: 80). The Puzrish-Dagan administrative
texts indicate that messengers from cities in the north received gifts. It has been argued that the
number of attestations of messengers from a particular place directly reflects the importance of
that place for the Ur III state. In that case, Mari, Ebla, Shimanum (near Diyarbakır?), and Urshu
(Samsat Höyük?) were the most important cities, followed by Khabura, Ninua, Talmush, Urkesh,
and Yamudium. Tuttul/Tell Bi’a is also mentioned, albeit only twice, indicating that it was of even
lesser political importance to the Ur III state than the cities mentioned above (Sallaberger 2007:
439). As far as can be judged from the Ur III documentation, no city in northern Mesopotamia
seems to have politically controlled large territories at this time (van de Mieroop 2007: 80).
The period from c. 2000 to c. 1600 bc was characterized by competing city-states that had varying success in subduing their rivals and incorporating them in larger political entities. Many of
these city-states were ruled by dynasties who claimed Amorite descent. Among the more powerful
city-states of this period were Yamhad and Qatna in the west, Assur in the northeast, Mari along
the Middle Euphrates, and Isin, Larsa, Eshnunna, and eventually Babylon in the south (van de
Mieroop 2007). However, the transition from the third to the second millennium remains poorly
understood, leading Charpin and Ziegler to describe it as ‘la grande vague’ (2003: 29). A clear view
of northern Mesopotamian political affairs only re-emerges during the period covered by the Mari
archives (Fig. 2.10).
During the Ur III period and probably until the end of the nineteenth century, Mari was ruled
by shakkanakku or governors. Although previously a brief hiatus was reconstructed between the
last shakkanakku and the first Amorite ruler (Durand 1985), recent epigraphical and archaeological research has provided evidence for a considerable degree of continuity (Butterlin 2007: 2301). The first Amorite ruler was Yagid-Lim, but not much is known of his reign and he was soon
succeeded by Yahdun-Lim (1810–1794 bc). Under Yahdun-Lim, Mari controlled the Euphrates
northwards up to Tuttul/Tell Bi’a, and also dominated parts of the Upper Khabur region, leading
to conflicts with Shamshi-Adad of Ekallatum (c. 1830–1776 bc). The border between the influence spheres of Yahdun-Lim and Shamshi-Adad may have been formed by the Wadi Jaghjagh.
During his reign, Yahdun-Lim also fought against a coalition of Yaminite rulers, which he defeated
at the Euphrates near Tuttul. Eventually, Shamshi-Adad defeated Mari and installed his youngest
31
Urkesh
Shubat-Enlil
Harran
Zalm a qum
Carchemish
A pu m
Ida- Ma ras h
Kahat
Nagar
Zalpah?
Kurda?
Num ha
Emar
Tuttul
A
Yamhad
Ebla
Qatna
Terqa
Assur
h
P
Eshnunna
Babylon
Isin
Larsa
u
ra
Mari
Fig. 2.10: Map of northern Mesopotamia showing early second millennium
study.
bc
tt
im
2000+ m
1500-2000 m
1000-1500 m
500-1000 m
0-500 m
0 km
intermittent lake
lake
N
30
60
names of sites and regions mentioned in this
son Yasmah-Addu as viceroy, while establishing himself at Shubat-Enlil/Tell Leilan and leaving
his eldest son Ishme-Dagan to rule Ekallatum. After Shamshi-Adad’s death, Zimri-Lim replaced
Yasmah-Addu as ruler of Mari. In the Upper Khabur, a kingdom of Apum appeared, centred on
Tell Leilan. Zimri-Lim was in turn defeated by Hammurabi of Babylon, who destroyed Mari in
1761 bc. A new middle Euphrates state emerged at Terqa, but that kingdom of Hana failed to exert an influence over other regions comparable to the authority of Mari (Charpin 1993; Charpin
and Ziegler 2003).
The political history of northern Mesopotamia during the third and early second millennium
bc reflects the conflicts between local city-states and foreign powers wishing to exert influence
over the region. The Akkadian presence in northern Mesopotamia, despite the fact that it remains
poorly understood, signified a first attempt at creating a larger regional state, followed by the
Ur III and Shamshi-Adad’s northern Mesopotamian states. Each attempt proved unsuccessful in
the long term, and especially the degree of administrative centralization that the Akkadian state
achieved in northern Mesopotamia may have been rather low. The Ur III state was highly bureaucratized, but again relatively short-lived and it did not exert clear permanent influence on northern
Mesopotamia. Zimri-Lim, on the other hand, seems to have exerted his influence by maintaining
alliances with local city-states, rather than attempting to control them directly through force.
32
3 Theorizing social responses to environmental change
3.1 Introduction
Section 2.3.7 has shown that the late third and early second millennium bc were characterized by a
climate that became progressively drier. Evidence from plants and human bone material suggests
that this aridification was associated with periods of environmental and nutritional stress, probably caused by insufficient water and/or low nutritional value of food intake (see Section 2.3.8).
Although the evidence for drought is currently relatively meagre, it nevertheless seems that the late
third millennium was a period during which shortages in water and/or food occurred more often
than before or afterwards, and that droughts must have been a regular feature of human life. The
present chapter will explore how social units react to shortage and more specifically how social relations develop under conditions of environmental stress. Section 1.3 laid down the requirements
that a framework for that type of study must meet. These requirements are (a) the ability of the
framework to explicitly consider the historical development of the position of human communities in their environment, (b) the need for a feedback mechanism between environmental change
and socio-cultural change, and (c) the consideration of communities, or possibly even households
as the fundamental social units that respond to environmental change, rather than societies as
a whole, even though these small-scale units cannot always be discerned in the archaeological
record.
This chapter aims to present and discuss a framework capable of meeting these requirements.
It will start with a delineation of the social scale at which social responses to environmental change
will be studied (Section 3.2). The following sections discuss the organization of responses to environmental change, and the maintenance of alternative responses (Sections 3.3 and 3.4). Section
1.4 already introduced the two hypotheses regarding the interaction between sedentary agricultural
and (semi-)nomadic pastoral communities: (a) it is to be expected that, as a result of environmental
stress, social relations between sedentary agricultural communities will develop along competitive
lines in order to improve each communities’ chance of survival and access to scarce and therefore contested resources, and (b) sedentary and (semi-)nomadic groups will develop cooperative
mechanisms ensuring mutual survival and maximization of their respective subsistence strategies.
This chapter will explore the logic of these hypotheses, and discuss archaeological correlates to
test them (Sections 3.5 and 3.6).
3.2 Defining the scale of analysis
It has been argued that a society is not an appropriate scale for the analysis of responses to environmental change, because a society rarely, if ever, acts as one single social unit (Section 1.3).
Below the level of the society, there is a potentially large range of scales at which an analysis can
be carried out. Following Binford, who argued that ‘a researcher’s choice of units and scales should
be determined by the potential for knowledge growth inherent in the alternatives’ (2001: 34), it is
here suggested that research at the scale of human societies should be abandoned in favour of a
more fine-grained perspective focusing on small-scale responses and decision-making. The present
study must be understood as part of a nested scalar approach to social responses to environmental
change. Within this scalar approach, there is not one legitimate approach at one appropriate scale,
but rather numerous approaches at different scales that complement each other (cf. Rosen and
Rosen 2001).
33
Examples of social units that may be taken as the basic unit of analysis include the individual
actor, the household, the community, or the state. These units can be placed on a scale of increasing demographic or spatial size, and, more importantly, on a scale of increasingly larger decisionmaking levels. Examples of this approach can be found in Halstead and O’Shea 1989b, Tainter and
Tainter 1996, and Wilkinson et al. 2007a. Alternatively, it is possible to focus on specific groups
ranked along a social axis within communities of a given scale. This is the approach adopted by
Rosen (1995; 2007) in her case-study on the responses to environmental stress by Early Bronze
Age urban elites. These two approaches are not contradictory, but highlight different aspects of
decision-making and social interaction. Rosen focused on decision-making within a social unit of a
given scale, arguing that the responses of Early Bronze Age urban elites were severely constrained
by social factors. The other approach focuses more strongly on interaction between independent
social units, rather than on internal social processes. Thus, this approach is more strongly concerned with external social processes, and how these are influenced by environmental change. The
internal and external approaches should therefore be considered complementary.
The present study takes the external approach and adopts the perspective of the settlement
community. This choice is partly determined by the history of research into late third–early second millennium bc climate change, and partly by the nature of the evidence. Many studies have
described the effects of late third millennium bc climate change at least partially in terms of settlement abandonment, that is, the replacement of an urban settlement system by a system characterized by fewer, and smaller sites. Although it has already been pointed out that settlement abandonment was less wide-spread than previously argued (Section 1.2), it is often still implicitly assumed
that settlement abandonment logically and automatically follows upon environmental deterioration. This study hopes to break away from this implicit assumption by explicitly theorizing how
settled communities can respond to environmental change. A second reason for adopting this scale
of analysis can be found in the available evidence. The settlement record of northern Mesopotamia
is relatively good and very diverse, offering good opportunities for detecting any social effects of
environmental change. Furthermore, the settled community as unit of analysis seems to coincide
relatively well with the available evidence on the political fragmentation of northern Mesopotamia.
Given the fact that highly centralized states were absent in northern Mesopotamia during much of
the third and early second millennium bc (see Section 2.5), it may be safely assumed that individual
cities maintained a relatively large degree of independence, and that interaction between them was
reasonably unrestricted by higher-order political entities.
3.3 Ordering social responses to environmental change
Every environment presents its inhabitants with risks due to unpredictability and patchiness in the
availability of resources. Successful existence in any environment is not based on its inhabitants’
abilities to cope with average environmental conditions, but on their ability to cope with extreme
variations in these conditions. Decisions will not be based on average conditions, but on the possible occurrence of potentially harmful extreme variations. Social units will thus develop behavioural strategies to cope with unpredictable periods of resource shortage (Winterhalder 1980:
147; Halstead and O’Shea 1989a: 1). Numerous ethnographic and archaeological studies testify to
the plethora of behavioural coping strategies that can be found among human groups. Observed
strategies include diversification and/or intensification of subsistence strategies, group fissioning, adoption of new resources, adoption of new mobility patterns, resource storage, exchange,
and the formation of social networks (e.g. Halstead and O’Shea 1989b; Tainter and Tainter 1996;
Winterhalder et al. 1999: tables 3, 4; McIntosh et al. 2000b).
The variability of mechanisms to cope with environmental stress is increased by the fact that
no correlation seems to exist between specific types of stress and specific responses. A review of
literature on responses to stress suggests that similar conditions of environmental stress have led
to a variety of responses among different groups and in different areas and periods (cf. studies
cited by Winterhalder et al. 1999). Even within groups multiple responses can be adopted due to
differences in, for example, wealth and availability of resources (Rosen 2007: 147). However, this
34
Theorizing social responses to environmental change
does not mean that these coping mechanisms were randomly adopted by groups affected by environmental stress. On the contrary, most researchers agree that the adoption of specific adaptive
mechanisms is a conscious choice and that the cost of the adaptation should match the severity
of the environmental change. In other words, extremely severe or persistent periods of environmental stress will lead to the adoption of increasingly more costly or disruptive adaptation mechanisms in terms of fitness, energy expenditure, effects on social status, or a combination of these
variables (Bates and Lees 1977: 838–9; Braun and Plog 1982: 508; Halstead and O’Shea 1989a: 5;
Minnis 1996: 67).
This approach assumes that it is impossible to continuously maintain buffering mechanisms
that are able to cope with every possible type of environmental stress. Maintaining such a general
mechanism would not only be extremely expensive, but would also require that environmental
stress can always be anticipated and/or predicted, which is unlikely to have been the case in any
past society. For this reason, it is much more cost-effective to organize buffering mechanisms and
adaptation strategies according to a hierarchy, whereby higher-level strategies are only adopted if
the strategies below them prove insufficient to deal with the resource shortage being experienced
(Tainter 1988: 110; Forbes 1989: 90).
Several researchers have proposed a correlation between the magnitude of resource stress and
the social cost of the adopted response (Braun and Plog 1982; Minnis 1996). They suggested that
increasingly severe periods of resource stress require the involvement of increasingly larger groups
of people, leading to stronger social obligations. In an ethnographic study on responses to subsistence crises in a mid to late twentieth century ad Greek farming community, Forbes suggested
that there was a distinction between low-level and high-level responses to environmental stress,
which he called Hazard-Response Mechanisms (HRMs). Low-level responses dealt with regularly
recurring short-term variations and were characterized by (a) continuous or frequent operation,
(b) high cost in terms of energy expenditure, (c) high integration in socio-cultural practice, and (d)
low visibility to those who are involved and to the observer. High-level responses, on the other
hand, dealt with less frequent but more severe resource shortages and were characterized by (a) infrequent operation, (b) low energy expenditure, (c) high social expense because normal social rules
are breached, and (d) high visibility. Examples of low-level responses observed in the Greek community were storage, diversification of crops, and scattering of farming plots across the landscape,
whereas high-level responses included begging and marrying off daughters to partners that would
be considered unsuitable under normal conditions (Forbes 1989: 95).
According to this model, low-level adaptation mechanisms create a safety-net against the possibility that environmental stress will result in resource shortages. These safety-net mechanisms are
maintained even though the cost in terms of energy expenditure is much higher than required for
immediate subsistence needs. By investing in this safety-net in times of abundance, it is hoped that
socio-culturally less-acceptable responses can be avoided when a period of environmental stress
actually occurs. For example, Forbes observed that there was a tendency to sow higher amounts of
grain and to store more foodstuffs than immediately needed, even if this meant that their market
price would go down as a result of rotting. However, this behaviour would ensure the presence of
foodstuffs in the case of an unanticipated shortfall in resource availability, such as a bad harvest
(Forbes 1989: 93). Similarly, Winterhalder et al. were able to show that Peruvian farming households scattered their agricultural plots to a degree that effectively allowed them to eliminate the
risk of shortfalls, even though this scattering led to a considerable reduction in the net production
that could potentially have been achieved by these households if they had consolidated their fields
(Winterhalder et al. 1999: 333–4). Thus, low-level responses lower the physical/nutritional as well
as social effects of environmental stress through continuous or regular investment in buffering
mechanisms against resource shortages.
High-level adaptation mechanisms, on the other hand, are only irregularly called upon; mainly
if low-level responses prove inadequate. It is important, however, that these responses do not necessarily result in permanent social change. According to Forbes, the high-level responses that he
observed in the Greek farming community ‘can be recognised not simply as social breakdown, but
as culturally patterned hard choices in the face of life-threatening hazard, for which lower-order
HRMs are inadequate’ (1989: 95). These high-level responses are, then, clearly within the recog-
35
nized spectrum of available responses. Extreme resource shortage therefore does not necessarily
lead to the adoption of entirely new responses, but to the acceptance of already known responses
that would otherwise be neglected in favour of less costly alternatives. In other words, an increase
in the severity of a resource shortage will not necessarily lead to an increase in the range of available alternative responses, but to an increase in the range of accepted alternative responses.
3.4 Maintaining alternative social responses to environmental change
If social responses to environmental change are hierarchically ordered, with high-level responses
only being adopted if low-level responses fail, it is likely that these high-level mechanisms will
be only infrequently called upon. Because such high-level responses are only irregularly used, it
is necessary to embed knowledge about these adaptation mechanisms in social practice to ensure
their continued existence beyond immediate need. McIntosh et al. refer to the concept of social
memory to describe mechanisms that aim to preserve vital knowledge of climates and environments of the past in societies that do not have recourse to written records (2000a: 24). Similarly,
Salzman (1978a) referred to the ‘cultural maintenance of organisational alternatives’ that could
be activated during periods when normal social structures failed. Through these mechanisms, behavioural alternatives can be maintained that can be activated when low-level responses to environmental stress prove insufficient. The operation and effectiveness of these mechanisms will be
discussed below.
A clear example of maintaining knowledge of behavioural alternatives was provided by Minnis
(1996: 62) in his discussion of adaptation strategies to resource shortage in the North American
Southwest, a region that experienced severe climate changes in the past. He drew attention to the
use of famine foods under conditions of environmental stress. Famine foods are foodstuffs that
are edible, but that are usually avoided because of their low dietary value or taste, unless there is
no alternative. Due to the fact that entire generations can come and go without ever experiencing
a crisis requiring the consumption of famine foods, the knowledge of these foodstuffs must be
passed on in the absence of actual experience. Minnis observed that in one case this knowledge
was maintained through the listing of edible plants in prayers, and through the annual ritual obligation to collect these plants. In this way, it was assured that knowledge of both the plants and
their growing locations would be maintained through time and passed on to later generations. The
recent decline in this knowledge can be attributed to changes in the subsistence base, and to the
inclusion of the region into the wider state and world economy.
A slightly different system was observed among Bedouin living in the Egyptian Eastern Desert.
There, during prolonged droughts, the Bedouin had to rely on drought-enduring plant species such
as the acacia tree (Hobbs 1990: 39). In fact, through various social mechanisms, the Bedouin even
ensured that these trees were protected from over-exploitation during periods without drought.
Thus, the Bedouin continuously maintained a religiously motivated prohibition on the cutting of
trees, they declared certain areas ‘lineage reserves’, falling under the protection of an important
leader, and they protected individual trees by marking them with a clan’s signature, thereby creating a sense of ownership (Hobbs 1990: 105–6). Similar regulations were also maintained for some
other shrubs, as well as certain animal species.
The concept of maintaining behavioural or organizational alternatives can not only be applied
to single adaptive responses such as famine foods, but also to more broad social structures that
strongly affect the way in which social units function and interact with others. Salzman has pointed
to the existence of ‘social structures in reserve’ that can be activated in periods of crisis or severe
environmental change (1978a, b). He suggests that such social structures in reserve can especially
be expected in ‘environments in which there is an alternation of conditions through time, in which
changes of circumstance are repetitive and to a degree of the same character’, but at the same time
points out that maintaining such alternatives provides ‘flexibility in adapting to new conditions’
(Salzman 1978a: 619). Salzman suggests that several systems exist through which social structures
in reserve can be maintained (1978a: 619–21): (a) the deviant minority, (b) operational generalization, and (c) the asserted ideology. Under the first mechanism, there exists a small group adhering
36
to (social) practices that are considered to be inappropriate under normal circumstances, but who
are nevertheless tolerated by the society at large. In this way, the minority maintains knowledge of
alternative behaviours, to which the society at large can revert in times of crisis. Under a system of
operational generalization, a social unit adheres to any of a number of organizational alternatives,
whereby the others remain within the ‘conceptual repertoire’ of the social unit. Salzman suggests
that this system is especially relevant for the maintenance of alternatives to short-term activities.
Finally, the asserted ideology represents a system under which social units acknowledge the existence and validity of a given social structure, but to which the social units do not actually adhere.
However, when faced with circumstances for which the asserted ideology is considered appropriate, social units will put that social structure into place and act accordingly.
In a comparative study on drought responses among pastoralist groups in Kenya and the
Negev, Bruins et al. (2003b: 29) documented the historical awareness or social memory of these
groups. They found that both the Maasai in Kenya and the Bedouin in the Negev remembered
years with severe droughts, and even assigned names to them. The Maasai used a system of naming that provided information on the severity of the drought. Thus, the word olameyu was used to
indicate drought in general, without any reference to its effects. The words emboot and emperi, on the
other hand, described droughts with particular emphasis on the adverse effects of droughts. An
emboot was a severe drought resulting in loss of livestock due to lack of pasture. An emperi drought
was even more severe and affected not only livestock but also endangered the lives of the Maasai
themselves. Such a drought could last over a year, resulting in mass migrations. Years with particularly severe droughts were remembered as far back as the 1920s. However, the study also indicated
that the word emperi, signifying the most severe droughts, was less well known among younger
Maasai because modern technology and international food programmes had reduced the impacts
of droughts. The Negev Bedouin employed a similar system of naming particularly dry years. This
system retained memories of dry years dating back to 1914.20
The Maasai and Bedouin cases are not strictly concerned with maintaining knowledge about
alternative response mechanisms. However, they do illustrate another important point, namely that
knowledge about past climatic fluctuations adds to the successfulness with which new situations
can be adapted to. Awareness of the potential variation of climatic fluctuations, even though they
cannot be predicted precisely, can help social units to maintain behavioural or organizational alternatives suited to these situations (Binford 2001: 42).
3.5 Modelling social responses to environmental change
Many responses to environmental change can be carried out by individual social units. Thus, an
individual household can diversify its subsistence strategies, intensify its production, or store resources for anticipated periods of shortage. In this sense, exchange and the formation of social
networks stand apart from other coping strategies because they require the involvement of multiple social units, even though cooperation between social units is regularly observed in the employment of other coping strategies. Similarly, mobility can also be considered a social strategy,
especially when the destination area of a relocating social unit is already occupied. Because the
present study is particularly interested in the emergence and development of such social responses
to environmental change, the interdependence of social structures and environmental change will
be discussed in more detail.
Social structures can serve to regulate a social unit’s access to scarce resources. This function
can be the result of the fact that a resource does not occur in the region inhabited by a social
unit, or because a social unit does not produce or extract that resource. Thus, social structures
20 It was observed in all regions that the knowledge about past resource shortages had declined significantly in the
recent past. In the North American Southwest, this decline resulted from changes in the subsistence base and from
the inclusion of the region into the wider state and world economy (Minnis 1996: 63). In the case of the Maasai and
the Bedouin, the negative effects of droughts are today more easily countered through better long-distance transport
options and the introduction of a monetary economy, allowing additional foodstuffs to be bought on the open market
(Bruins et al. 2003b: 34). The waning importance of the term emperi in the Maasai case illustrates how this memory
evolves under changing social conditions, again emphasizing the social aspect of environmental stress.
37
can reduce the effects of environmental variability through their ability to forge relations between
social units with differentiated access to resources (Rautman 1996: 199). Differentiated access to
resources is usually linked to the geographical space that a social unit occupies or uses, and to its
(in)ability to exclude others from using that area or resource. Under these conditions, various social structures can emerge that regulate access to scarce resources. Depending on socio-cultural
conditions and the nature of the resources that are involved, these structures can include exchange
networks, hierarchical relations between groups, territoriality, and conflict (Dyson-Hudson and
Smith 1978; Boone 1992; Kelly 1995: 161; Kohler and Van West 1996).
Which social structure regulating access to resources will emerge primarily depends on the resources that are utilized by each group within the network, on the nature of resource availability
for each group, and on differences in resource availability between groups. These variables can be
expressed along two axes as inter-group correlation of resource availability and intra-group variation in resource availability, resulting in four different configurations of related social structures
regulating access to resources (Table 3.1). Inter-group correlation indicates the degree to which
groups experience resource shortages and abundances at the same time, with a high correlation indicating strong temporal synchronization of resource fluctuations, and a low correlation indicating
that groups usually do not experience resource shortages at the same time. Intra-group variation
indicates the degree to which a group is likely to experience resource shortage, with high variation
indicating that there is a high probability of resource shortage, but also the potential for producing
a surplus. Low variation indicates that the overall production of a group remains stable through
time. In other words, production is relatively predictable (Kelly 1995: 198). 21
It is clear from Table 3.1 that the social structures are largely determined by inter-group correlation, with intra-group variation in resource availability mainly affecting the intensity of the
structures. Thus, spatial boundary defence, territoriality, and conflict are more strongly associated
with high correlation of resource availability between groups. This situation is represented in configurations (A) and (C). Social boundary defence and exchange, on the other hand, are the main
social structures in configurations (B) and (D). Physical storage of resources, on the other hand,
seems more strongly associated with intra-group variation and is expected in configurations (A)
and (B). These associations point to a fundamental difference in how groups interact with others
when faced with resource shortage. This difference can be summarized as the decision by individual social units whether or not it is possible or advantageous to share/exchange resources with
other groups.
If neighbouring groups experience synchronous shifts in resource availability, that is intergroup resource availability is highly correlated, resource shortfalls and periods of abundance will
occur at the same time. During a period of abundance, every group has sufficient resources and
therefore does not need to procure them through exchange, whereas during a period of shortage,
exchange is likely to fail because no group has resources to share or exchange. Thus, if social units
high inter-group correlation
of resource availability
low inter-group correlation
of resource availability
high intra-group variation
in resource availability
(A) -household storage
-restricted social access between groups
-spatial boundary defence
-strong territoriality
-conflicts common
(B) -social boundary defence
-some storage
-intensive and rigid reciprocity between groups
low intra-group variation
in resource availability
(C) -passive territoriality
-long-distance migration
-conflicts rare
(D) -relaxed social boundary defence
-differentiated exchange between groups
Table 3.1: Expected social structures regulating access to resources under different conditions of resource availability (based
on Kelly 1995: fig. 5-6).
21 Although the configurations in Table 3.1 were originally proposed for hunter-gatherer societies, much research on
interaction between more complex agricultural and/or pastoral communities has actually been based on the same
underlying principles, as is for example the case with the work by Cashdan (1987) and Rautman (1996), both discussed
in Section 3.5.2. See also Wendrich and Barnard 2008 on commonalities between research on hunter/gatherers and
pastoralist and even settled agriculturalist societies.
38
with similar subsistence strategies experience similar conditions of environmental stress, they will
have less incentive to engage in exchange networks with each other because it is less likely that
goods can be reciprocated. Conversely, when groups rely on different resources, or when periods
of resource availability are asynchronous, that is the correlation of inter-group resource availability
is low, there is much more incentive for exchange, simply because resources are available. Below,
these two response configurations will be explored in more detail.
3.5.1 Territoriality, social hierarchy, and conflict
Territoriality is the restriction of access to resources by one social unit and can be achieved
through social regulation or through spatial exclusion, known as social boundary defence and
spatial boundary defence, respectively (Casimir 1992). In the previous section, it was suggested
that territoriality, conflict, and spatial boundary defence are expected adaptive strategies against
resource strategies among groups experiencing synchronous fluctuations in resource availability.
This synchronicity will often be the result of a reliance on similar resources, or because they are
engaged in similar subsistence strategies. Even under these conditions, however, territorial behaviour will not always emerge. The concept of economic defensibility has been used to explain the
occurrence of territorial behaviour: ‘territorial behavior is expected when the costs of exclusive
use and defence of an area are outweighed by the benefits gained from this pattern of resource
utilization’ (Dyson-Hudson and Smith 1978: 23). Economic defensibility is a function of the predictability and density at which the competed resource occurs, and the time and energy that is
needed to defend it. Thus, in terms of the availability of resources, territorial behaviour is most
likely to occur when resources are dense and predictable. Furthermore, territoriality is likely to increase together with investments that are made in a competed resource (Stone 1994). In any other
constellation, other behaviour may be more appropriate to regulate access to resources (DysonHudson and Smith 1978: 25).22
The choice for conflict is based on an assessment of the risks involved, and of those inherent
in the alternatives. Given the high risks associated with conflict, it is likely that social units will
first explore alternative methods of gaining access to the competed resource. Instead of open
conflict, social units seeking access to a resource can accept a lower social status with respect to
the social unit controlling the resource, or a social unit can move away. A social hierarchy is only
likely to emerge if the social unit controlling the contested resource benefits from the acceptance
of subordinate social units, and if social units seeking access to the competed resource are willing
to accept a lower position (Boone 1992: 312). Movement is only possible if the resource can be
obtained elsewhere (Carneiro 1970: 735). If these requirements are not met, conflict becomes an
economically advantageous and sometimes unavoidable strategy to gain access to a resource. In
this sense, violence and warfare are not inherently different from alternative methods of resource
procurement because the decision to engage in conflict to gain access to resources derives from
the same assessment of risks and opportunities as would have alternative responses.
The correlation between conflict and resource availability was investigated by Ember and
Ember (1992). They conducted a cross-cultural ethnographic study on the frequency and causes
of warfare. Their aim was to investigate whether there was a correlation between warfare and deteriorating environmental conditions. The underlying idea was that resource scarcity, or unequal
access to resources, would lead to competition. Ember and Ember used the Human Resource Area
Files (HRAF) ethnographic sample to test this hypothesis. They analysed a sample of 186 mainly
non-state societies and looked for a correlation between warfare and a number of socio-cultural
variables including threat of natural disasters, threat of famine, chronic resource scarcity, and several social conditions that were often assumed to increase the potential for warfare. The threat of
natural disasters and famine was argued to be present when the ethnographic literature indicated
that societies feared a disaster or famine, even though these events had not taken place during their
recent past. This analysis indicated that fear of nature, that is, the perceived possibility that some
22 For example, Dyson-Hudson and Smith (1978: 25) used the term superabundance to describe situations in which a
resource is so abundant that territoriality is economically and energetically unviable, so that allowing others to use the
resource freely is the best solution (cf. Barnard 1992: fig. 4.1).
39
natural disaster or famine would cause loss of vital resources, and chronic resource scarcity were
good predictors of warfare. Further analysis revealed that fear of a natural disaster was probably
also a cause of warfare. In those societies where the ethnographic studies revealed a fear of natural
disasters, conflict and warfare were often endemic and resulted in the capturing of resources, even
when there was no shortage at that time. It seems, then, that in many societies warfare may indeed
have been actively used as a buffering strategy against unpredictable resource scarcity (Ember and
Ember 1992: 257).
The processes that are involved in the emergence of territoriality and conflict are likely to be
intensified by demographic pressure on resources. If the number of people relying on a given
resource increases, the per-capita return decreases and eventually a group may decide to actively
resist growth and exchange, resulting in an increased potential for conflict (Zhang et al. 2007:
19218). The complex interaction between resource use and the incentive for exchange between
social units with similar subsistence strategies has been modelled by Kohler and Van West (1996).
According to them, social units base their decision on whether to cooperate with others and share/
exchange a resource, or to act individually, on the risk associated with obtaining that resource. If
the average production of a group is below the subsistence needs of each individual group member, but a member can potentially produce sufficient resources for himself due to unpredictable
variability, that member is better off acting independently. If a member acts independently, there
is no risk that he has to share or exchange his potentially above-average production with underproducing group-members. Conversely, if average group production is above the subsistence needs
of individual members, group members will likely remain within the group because any unpredictable shortfalls can be compensated by the production from other group members. Thus, if social
units with similar subsistence strategies experience similar conditions of environmental stress, they
will have less incentive to engage in exchange networks with each other because it is less likely that
goods can be reciprocated. Even though sanctions on free-riding will likely exist, that is profiting
from a system or network without investing in it, such sanctions will become increasingly strained
and may eventually break down, resulting in competition, territoriality, and conflict.
3.5.2 Exchange networks and social structure
Exchange as a buffering mechanism against resource shortage is especially expected among groups
experiencing asynchronous fluctuations in resource availability. Exchange networks function as
buffering mechanisms against resource shortage because they possess a spatial as well as temporal component (Minc and Smith 1989: 65; Minnis 1996: 9). Through their spatial component, exchange networks allow the distribution of resources from areas where they are widely available to
areas of scarcity. The temporal component of exchange networks allows participants to acquire resources without directly reciprocating. In other words, there can be a time-lag between the acquirement of resources and their payment. Especially the temporal component of exchange networks is
likely to be influenced by the social relations existing between the participating social units because
delayed reciprocity requires social units to trust each other that an exchange will be reciprocated at
a later point in time. Manipulation of social relations between social units can therefore affect the
nature of exchange taking place between these units.
Whereas Braun and Plog argued that ‘sustained increases in the intensity of integration will
occur as concomitants of sustained increases in region-wide uncertainty or risks arising from environmental change’ (1982: 508), Rautman (1996) proposed a more specific correlation between
environmental change and the emergence or reorganization of exchange networks. She suggested
that exchange networks will preferentially emerge between locations that are not equally vulnerable to environmental change (Rautman 1996: 201; Whallon 2006: fig. 2). This point of view coincides with configurations (B) and (D) of Table 3.1, characterized by high inter-group variance in
resource access. Thus, if regions experienced differences in environmental variability, it might be
expected that exchange networks will develop between these regions, or that existing networks are
reinforced or strengthened. Similarly, Read and LeBlanc have shown through theoretical modelling
how two groups experiencing different levels of resource availability through time and space can
40
attain a higher average production over time through cooperation and therefore have a competitive advantage over groups that experience one of these levels of resource availability exclusively
(2003: 72).
However, Rautman assumed that a similar subsistence economy existed in each region, based
on the same resources. Under such a scenario, shortages in one region could only be alleviated
through imports from elsewhere. If in one region multiple subsistence economies relying on different resources existed, it is likely that social networks will emerge between social units participating in different economical systems. This means that if social units experience environmental
stress, they will preferably establish or invest in social relations either with social units that live in
regions where there is less environmental stress, or with social units that pursue different subsistence strategies (cf. Kelly 1995: 200–1). Johnson and Earle point to the similarity between the concepts of economy and niche as expressions of subsistence specializations within human societies
(2000: 22), thereby reinforcing the distinct nature of each strategy and the potential for exchange
between their practitioners. Cashdan (1987) has shown that economic specialization within a limited region indeed leads to stronger exchange relations between specialized groups than between
groups practising more generalized strategies.
It has long been recognized that exchange networks are often built on, and couched in the terminology of social structures such as kin, gender, age, class, ethnicity, and ancestry. Depending
on the determining structure and the scale of the network, real, perceived, or constructed shared
identities between the engaged social units can exist. For that reason, exchange networks and social
structures as described above are related concepts that can be difficult to disentangle. This interdependence between exchange networks and other social networks can be described through the
concept of reciprocity. Reciprocity is the exchange of goods and/or services, whereby giving creates an obligation to return goods of equal value by the receiving party. Reciprocity can be ranked
from generalized through balanced to negative reciprocity, whereby the need to reciprocate goods
with equal value and directly at the time of the actual exchange increases from generalized to balanced reciprocity. Generalized reciprocity is characteristic of households or small communities,
with reciprocity becoming increasingly balanced as social distance between social units increases.
Negative reciprocity includes theft and raiding (Sahlins 1972: fig. 5.1; Johnson and Earle 2000:
47–50).23
Several researchers have investigated the correlation between environmental change and the
development of exchange networks and social structures (e.g. Braun and Plog 1982; Minc and
Smith 1989; Rautman 1996; Rowe 2002). For example, Nettle (1996; 1998) found a correlation
between language diversity and environmental risk. An analysis of the distribution of languages
across the globe suggested that the number of languages was negatively correlated with the risk of
crop failure. In areas with a low risk of crop failure, the number of languages was high, whereas
in areas with a high risk of crop failure, the number of languages was low. Because languages play
a fundamental role in communication, their distribution can serve as a proxy for social networks
existing between spatially separated communities. The negative correlation between language diversity and environmental risk therefore suggests that social networks are indeed partially shaped
by environmental conditions.24
Cashdan observed an even stronger correlation between environmental variability and the
emergence of social structures aimed at the creation of a social environment favourable to exchange (2001a). Using a global cross-cultural sample, Cashdan found that climatic variation and
predictability were among the strongest predictors of ethnic diversity. Ethnicity is here defined as
an identity that is recognized by the ethnic group itself, as well as by outsiders. Ethnicity is usually
based on a common ancestry and often includes perceived or constructed elements of kinship, a
common language, and a common past (see Section 7.3.1). In areas with an unpredictable environment, and more particularly unpredictable rainfall, ethnic diversity was low. These conditions also
23 Note that even generalized reciprocal exchanges are conducted to achieve personal gain. In fact, personal gain, however defined, is inherent in every reciprocal exchange; otherwise giving or sharing social units would run the risk of
being exploited by free-riders attempting to get goods without reciprocating (Boone 1992: 271).
24 Nettle, as well as Cashdan in the following example, compensated for the possibility that lower language diversity, as
well as ethnic diversity in Cashdan’s case-study, could have resulted from lower population densities.
41
led to higher intra-ethnic loyalty, that is, the degree to which group-members favour others within
the same ethnic group over outsiders (Cashdan 2001b: 761–2). Cashdan argued that this correlation can be explained by the role of ethnicity in ‘providing social insurance against unpredictable
risks’, resulting in the following generalization: ‘where environmental unpredictability requires
reliance on reciprocal or redistributive mechanisms among a spatially dispersed group, ethnic diversity should be lower’ (2001a: 979). Thus, ethnicity functioned as a shared identity aimed at creating a social environment in which reciprocity tends to be more balanced or even generalized (cf.
Horowitz 1985: 81; Bentley 1987: 42). This observation, then, strengthens the correlation between
language diversity and environmental variability.
3.5.3 Hypothesizing social responses to environmental change
Summarizing, the present discussion has shown that social structures play an important role in
reducing the effects of environmental change. More specifically, it is argued that social networks
aimed at allowing and regulating exchange will emerge or will be intensified between social units
experiencing different levels of environmental risk, either because they live in different environments, or because they employ different subsistence strategies. It is furthermore suggested that the
emergence of social networks rests on the willingness of social units to engage in exchange networks. Resources are unlikely to be exchanged if environmental stress among social units within
the exchange network happens simultaneously for each social unit. Furthermore, exchange is less
likely to take place if individual social units risk having to share with under-producing social units.
Under such circumstances, processes that are aimed at regulating access to resources will likely
emerge, to ensure that the controlling social unit can meet its subsistence needs when experiencing
environmental stress. Alternatively, conflict may arise or intensify.
Based on these considerations, it is now possible to formulate hypotheses regarding the formation, maintenance, and disintegration of social networks at various scales and under various environmental conditions. These hypotheses are the following: (a) under resource stress and within
similar environments, that is environments with a similar risk regime, competition over scarce
resources between social units will increase; and (b) under increasing resource stress, cooperation
will emerge between social units that experience different regimes of environmental risk, for example due to different environmental conditions, or as a result of different subsistence production strategies. Within the context of the present research focussing on social responses to late
third-early second millennium bc climate change in northern Mesopotamia, these hypotheses can
be rephrased as follows: (a) it is to be expected that, as a result of environmental stress, social relations between sedentary agricultural communities will develop along competitive lines in order
to improve each communities’ chance of survival and access to scarce and therefore contested resources, and (b) sedentary and (semi-)nomadic groups will develop cooperative mechanisms ensuring mutual survival and maximization of their respective subsistence strategies. The archaeological
correlates of each of these hypotheses will be discussed below.
3.6 Archaeological correlates of competition and cooperation
In the present study, two main proxy records will be employed to study the degrees of competition
and cooperation between social units: the settlement record of northern Mesopotamia, and the
historical record that provides details about the constellation of northern Mesopotamian society
during the third-early second millennium bc. The settlement record will serve as a proxy of competition and cooperation between sedentary communities, whereas the historical record will document the formation and reorganization of social relations between sedentary communities and
(semi-)nomadic communities. The choice for these records will be discussed below, as will several
additional proxies that can potentially support the argument made here.
Human dispersal across the landscape is not random; it reflects social structure as much as
it reflects the environment itself. Settlement systems are organized according to a number of
environmental and social principles (Roberts 1996: fig. 1.5). Depending on which principle is
given priority, a number of settlement systems can emerge, ranging from fully nucleated to fully
42
dispersed, and including everything in between (Roberts 1996: 19). Economic practicality often
dictates that households focusing on subsistence agriculture live on or next to their land in order
to reduce transport and travel costs, resulting in a dispersed settlement pattern. However, social
relations, obligations with other households, or safety reasons may lead households to live in nucleated communities or at least close together (Roberts 1996: 35; Peterson and Drennan 2005: 6).
Furthermore, environmental variables make certain places in the landscape more suitable for habitation than others, depending on for instance the availability of water, good soils for agriculture or
grazing, presence of game animals, or defensibility. In many cases, the choice for a particular site
will be the result of a compromise between these social and environmental variables.
It can therefore be assumed that settlement patterns are the result of conscious human behaviour, and that settlement analysis can be undertaken to gain insight into that behaviour.
Archaeologists have routinely employed various settlement pattern analyses to investigate the issues addressed here. Among the more well-known analyses are nearest neighbour analysis and
rank-size analysis (e.g. Johnson 1980; Meijer 1990; Ristvet 2005). In nearest neighbour analysis, the
distance between each settlement and the surrounding settlements is calculated, and these figures
can be used to calculate average, minimum, and maximum distances between settlements. Nearest
neighbour analysis can be used to detect settlement clusters, which in turn can be interpreted in
terms of social relations existing between nearby settlements. Rank-size analysis ranks sites according to their size, and the resulting graph can be used to posit interdependence between sites,
and to explore issues of population nucleation and dispersal.
Although nearest neighbour and rank-size analyses are clearly interrelated methods, they require different data in order to be effective. Nearest neighbour analysis assumes that distance between settlements is a meaningful socio-cultural category and in some way correlated to social relations and economic interdependence. Conventional nearest neighbour analysis calculates distances
between sites as if they were located on a flat, featureless plain. It is much more likely that the
actual travelling distances were a more socially meaningful category, but these economic distances
are difficult to calculate in the absence of detailed information on the past landscapes. The results
of nearest neighbour analysis should therefore be particularly suspect in unstable environments
such as river valleys, where shifting river beds may continuously eradicate and create opportunities
for movement.
Rank-size analysis assumes that the size of a settlement is a relative indication of the importance and position of that settlement in a larger settlement network or wider region. The necessary
data may be entirely drawn from archaeological sites, and may therefore be more useful in areas
with relatively unstable environments, although the processes that modify the landscape in general
may also affect individual sites. As will be argued in Section 4.2, the degree to which Early and
Middle Bronze Age sites were affected by post-depositional and environmental processes seems
modest when compared to other archaeological periods, suggesting that rank-size analysis indeed
has some value in evaluating settlement systems dating to the northern Mesopotamian Early and
Middle Bronze Ages. Chapter 4 will discuss in more detail the process of generating and interpreting rank-size plots, and Chapter 5 will present the data on which the analyses will be carried out.
Exchange networks serve to disperse goods and/or information across the landscape. As with
human settlement, this dispersal is not random and reflects conscious choices, needs, and opportunities of the participants. Such networks can be studied directly through sourcing a given material and identifying its distribution, or indirectly through the study of the distribution of containers within which perishable goods were transported and exchanged. The spatial distribution of
iconographic styles or motifs can be tentatively used as a proxy for social contacts. Similarly, the
distribution of luxury items is commonly argued to reflect social interactions (e.g. Whallon 2006).
Although these distributions highlight the existence of some dispersal mechanism, they do not
identify the processes by which materials moved across the landscape. These approaches, then,
are problematic for the social network that is studied here, particularly because the hypothesized
exchange goods, that is, wool, foodstuffs, and possibly information, leave almost no traces in the
archaeological record. A further problem is the fact that historical sources for the early second
43
millennium bc in northern Mesopotamia are generally not concerned with economy, including
exchanges between sedentary and (semi-)nomadic communities, although references to such interactions do occur in certain periods.
However, texts for the late third-early second millennium bc provide information on a different aspect of exchange networks, namely the social context in which such networks operate. This
context consists of the social identities of the participants, that is, those aspects of social identity
that are shared by each participant and that form the basis for reciprocity. In Section 3.6, it was
suggested that these identities can function as a proxy for the existence of exchange networks. It is
not argued that the existence of a shared identity is equal to the existence of an exchange network,
but that such networks will more easily emerge between social units sharing common identity
traits, and that these networks will be more effective in alleviating environmental stress. According
to this view, shared identities will be emphasized by social units participating in the same exchange
networks. Chapter 7 will discuss the formation and spatial distribution of shared identities in
northern Mesopotamia, and evaluate which social units were involved.
3.7 Conclusions
It can be argued that, despite the large variety of responses to environmental stress, these responses can be grouped according to one of two different principles: the associated social and/or
energetic cost or disruptiveness of a given response, and the nature of resource availability. The
first principle dictates that responses with a low cost will be adopted before responses that have a
higher cost or that are more disruptive. Low-level responses deal with regularly occurring smallscale shortages, whereas high level-responses deal with increasingly severe fluctuations, often at
an increasingly greater social cost or involving larger social groups. Whereas low-level responses
are strongly embedded in social practice, the infrequent operation of high-level responses requires
the existence of effective mechanisms for the transmission of knowledge about these responses
through periods when they are not required. Most importantly, high-level coping mechanisms
should not be considered as unique responses, but as socio-culturally structured behavioural alternatives for periods when the regular accepted range of strategies turns out to be inadequate.
The second ordering principle dictates that under specific conditions of resource availability,
certain social responses are more likely to emerge than others. Social responses to environmental
change, that is those responses involving multiple social units, are fundamentally concerned with
the decision whether or not it is advantageous to exchange resources with others. It is expected
that social units will exchange in resource networks if there are differences in resource availability.
If resource availability is synchronous across multiple social units, acting independently is a better
strategy to buffer against resource shortage, and territoriality and conflict are much more likely to
emerge.
44
4 Methods for the reconstruction of regional
settlement trends
4.1 Introduction
This chapter is concerned with methods for the reconstruction of regional demographic trends
through the study of the material record yielded by archaeological excavations and regional archaeological surveys. In archaeology, surveying is a type of research whereby an area is subjected to
surface inspection with the purpose of identifying archaeological remains. Through their regional
coverage, surveys complement the picture emerging from single, excavated sites. Surveys have thus
proven an invaluable tool for archaeologists, allowing them to address a whole range of research
questions that are beyond the scope of single site excavations. Surveys have been especially successful in highlighting demographic changes, settlement patterns, and landscape use in general
(Wilkinson 2000a; Banning 2002; Wilkinson 2003). Like excavated data, survey data can only be
interpreted through a large body of middle range theory. There is a lively debate on the limitations
of survey data, and how gaps in the survey record can be recognized and compensated for. The
question as to which problems can be addressed by survey data, and what their relation is with
other data sources, is less clear. Many of the inherent flaws of surveys can be compensated for
through the application of statistical methods, and by the use of survey data with as many other
sources as possible (Bintliff 2000: 214). Therefore, many researchers hold a positive view of survey data as a reliable source for the reconstruction of the past (e.g. Bintliff et al. 2000; Wilkinson
2000a; Banning 2002).
Since the early to middle twentieth century, archaeological surveys have been a vital part of
the toolkit of archaeologists working in greater Mesopotamia. Starting with the pioneering work
by archaeologists like Braidwood, Jacobsen, and Adams (Braidwood 1937; Jacobsen and Adams
1958; Adams 1965), a large number of systematic archaeological surveys has been carried out with
various underlying strategies and aims (see for a recent overview Wilkinson 2000a, 2003). Surveys
are no longer just a means to select a site suitable for excavation but have become a research field
in their own right, with a separate methodology and sometimes driven by very specific research
questions (Banning 2002).
The early survey work in the Jezirah is characterized by, but not limited to, the fieldwork of
Poidebard, Mallowan, and van Liere and Lauffray. Regional investigation of archaeological remains in the Jezirah started with the aerial reconnaissance by Poidebard (1934) in the 1920s and
early 1930s. He photographed numerous tells and landscape elements and created an aerial photographic record devoid of the modern overburden of agricultural and construction activities, that is
useful up to this day. In the 1930s Mallowan surveyed and excavated numerous sites in the Upper
Khabur and the Balikh Valley, including Brak and Chagar Bazar (Mallowan 1936, 1946). Using
French aerial photographs from the 1920–30s, van Liere and Lauffray recognized and mapped the
radial lines extending from many tells in the Upper Khabur (van Liere and Lauffray 1954).
Lower Mesopotamia has always been a major focus for fieldwork, but the political events of
the last decades have brought major changes, resulting in the switching of much research to Syria.
Thus, the 1960s and 1970s saw many surveys being conducted in southern Mesopotamia, whereas
in southeast Turkey and the Jezirah survey research with strict methodologies only really started
in the following decades, except for the Euphrates Valley, where dam construction necessitated
intensive archaeological research from the 1970s onward. The survey and excavation bias toward
easily accessible river valleys where the largest sites can be found, resulted in the one-sided view
45
that ancient Mesopotamian civilization, or North Syrian for that matter, principally focused on
these valleys, the Kranzhügel phenomenon being a notable exception. Recent surveys aimed to correct this bias.25
This chapter aims to discuss the strengths and weaknesses of survey evidence, suggesting
that statistical analysis can and indeed must fill in some of the gaps inherent in archaeological
survey data. It starts with an outline of the limitations and potential of survey data, focusing on
Near Eastern survey practices. This discussion focuses on three key issues: (a) the visibility of archaeological periods in surface artefact collections (Section 4.2), (b) the problems associated with
chronology construction (Section 4.3), and (c) the contemporaneity problem (Section 4.4). There
follows a discussion of the theory that is necessary to translate survey data into historical population levels. Three steps are considered here: (a) simulation of site contemporaneity, (b) estimation
of total occupied area, and (c) reconstruction of on-site population densities. It is argued here that
this procedure is suited to document trajectories of growth and decline, rather than individual settlement histories. Thus, it has a high comparative value, and it will be used for that purpose here.
Complementarily to this demographic reconstruction, rank-size analysis will be carried out on the
studied settlement systems in order to study their organizational aspects. Section 4.6 will discuss
the methodological aspects of rank-size analysis, and consider the various explanations that have
been offered for different rank-size distributions. In Chapter 5, the demographic and rank-size
analyses will be performed on datasets from three surveys in northern Mesopotamia, namely the
Birecik-Euphrates Dam Survey (B-EDS), the Balikh Survey (BS), and the North Jazira Survey
(NJS).
4.2 Interpreting survey evidence: limitations and possibilities
Straightforward interpretation of the results of archaeological surveys is impossible. Many studies
have highlighted the difficulties of interpreting archaeological surface collections in terms of site
numbers, sizes, and occupations of sites (e.g. Ammerman 1981; Redman 1982; Wilkinson 2000a;
Banning 2002). These difficulties stem partly from the acknowledgement that the composition of
particular surface collections is shaped by a large variety of socio-cultural processes and that surface collections as perceived by archaeologists are the result of very complex taphonomical processes. Any attempt to interpret archaeological surface collections must therefore take into account
the possibility that (a) the sites and landscape features that are recognized do not represent the
whole spectrum of sites and features that existed in the past; (b) the surface collection does not
reflect the true diversity of materials and periods that is present at a surveyed location; and (c) the
surface collection does not necessarily provide insight into the socio-cultural processes leading up
to the deposition.
Early surveys focused on tell-sites that are easily identified from the ground, often from the
seat of a car, and on aerial/satellite imagery and stand out as major foci of human activity. The
most important consequence of this strategy is that smaller, non-mounded sites, such as pastoral
campsites, will invariably be underrepresented. Several survey projects have included an off-site
transect-walking component to assess the degree to which the landscape between tells is filled with
archaeological remains (e.g. Wilkinson and Tucker 1995; Ur 2004; Wilkinson 2004). Apart from the
recognition of many non-habitation features, and in fact a continuous archaeological landscape,
such intensive surveys have also resulted in the recognition of low settlement sites that would not
have been found by less intensive surveys (Redman 1982: 377). A further bias toward large sites
can be introduced by the geomorphology of a region. In river valleys, alluvial sedimentation can
for example easily cover entire flat sites and even smaller tells. Heavy erosion may result in the
complete removal of entire occupation layers from the flanks of settlement mounds (Postgate
1994: 50). The inclusion of geomorphological research in survey methodology has allowed the
25 Examples are the Wadi Ağiğ survey (Bernbeck 1993), the surveys in the area between the Euphrates and Balikh Valleys
(Einwag 1993; Danti 1997), the Western Khabur survey (Hole 1997a), the survey of the Jebel Bishri to the south of
the Euphrates (Lönnqvist 2006), and the Wadi Hamar Survey around Tell Chuera and Kharab Sayyar (http://web.
uni-frankfurt.de/fb09/vorderasarch/survey.htm (Accessed 16 November, 2007)).
46
Methods for the reconstruction of regional settlement trends
reconstruction of anthropogenic influence on the landscape through time, and the evaluation of
problems of site preservation and visibility. A further step is the recognition of landscapes of survival and landscapes of destruction, resulting in an apparently uneven distribution and survival of
sites and periods across space (Wilkinson 2000a: 229).
It was recently noted that there is a large discrepancy between the results of field surveys conducted in Greece and Italy, and in the Near East (Wilkinson et al. 2004). A review of multiple recent surveys in both areas showed that the number of recorded sites per square kilometre for Near
Eastern surveys is far below that for modern surveys in the Mediterranean, mainly Greece. At first
sight, this discrepancy appeared to result from the relatively low intensity of Near Eastern surveys,
as compared to surveys elsewhere. However, comparison of the results and strategies of several
northern Mesopotamian surveys suggested that other factors than post-depositional processes or
survey biases may have been responsible for this discrepancy. Firstly, the northern Mesopotamian
surveys that included a fieldwalking component found relatively few sites that would not otherwise have been found with less intensive strategies. Secondly, investigation of sections resulting
from modern digging activities suggested that alluviation was not or barely obscuring sites outside river valleys. Finally, it was suggested that it is more efficient to make new mudbricks than
to reuse old ones, resulting in relatively minor destruction of old sites in the search for building
materials. These points suggest that a survey that combines intensive field-walking with satellite/
aerial imagery analysis can assemble a relatively good record of archaeological features in a region
(Wilkinson et al. 2004: 196).
On the level of the individual site, tell formation processes will likely cover early periods with
the overburden of later occupations. As a result, pottery from the earlier periods will be less represented in the surface assemblage, and any interpretation based on the number of sherds and
their occurrence on the surface may be biased. However, there is some reason to believe that this
overburdening may be less problematic for Early/Middle Bronze Age sites than for earlier periods.
Wilkinson (2000a) has plotted the ratio of occupations on tells versus occupations at low sites of
less than 5 m high through time for the Tell Beydar Survey area, and found that particularly the
third millennium bc represents a period during which settlement nucleated on high tells. After that
period, the significance of tell-based settlement decreased, in favour of lower and smaller sites.
This development suggests that third millennium bc occupations in particular will be highly visibly in the Jezirah. Evidence from rescue excavations in the Birecik-Euphrates Dam area suggests
that there, too, third millennium occupation is particularly well represented on tells and therefore
highly visible (see Section 5.2.1). The discrepancy of tell-based versus low site occupation is less
pronounced during the Middle Bronze Age, suggesting lower visibility for this period, but with the
caveat that overburden from later periods is also relatively limited. 26
There is also discussion about the degree to which the archaeological surface assemblage reflects the chronological components of a site. Surveys usually assume that there is a positive correlation between the surface assemblage and the archaeological contents of a site. Yet it cannot be
ascertained that all periods present in a tell will be equally represented on the surface. For example,
surface assemblages may be subjected to stronger erosion than buried artefacts, meaning that large
or high-fired sherds will dominate the record (Postgate 1994: 50). The preference of individual
surveyors to pick up highly visible sherds may leave certain unremarkable types underrepresented.
Furthermore, re-use of mudbricks, or construction of mudbricks from earth taken from tells, may
result in the displacement of archaeological material. However, research at sites that are surveyed
as well as excavated seems to suggest that there is some agreement between later ceramic periods
inferred from survey, and those actually excavated. Pre-pottery periods, on the other hand, often
go unnoticed and their presence can only be ascertained through excavation. Similarly, the on-site
distribution of sherds on the surface is correlated to past settlement and activity areas of the site.
26 A similar transition from nucleated, tell-based occupation to dispersed occupation on non-mounded sites has been
noted for the Amuq Valley, but there the transition occurred between the Iron Age and the Seleucid periods (Casana
2007: 203–4).
47
In both cases, the correlation is not very strong, but sufficient to use it as an indication of site size
(Lyonnet 2000: 7, cf. Redman and Watson 1970 and Flannery 1976, but see Kohlmeyer 1981 and
Schiffer 1987: 126, 353).
The results from archaeological surveys are thus nothing more than a sample. Just how representative this sample is, and how much of the total assemblage is actually recovered, are the main
issues that researchers have to address to correctly interpret their data. The comments made above
suggest that the site recovery rate for Near Eastern surveys can be relatively good, depending on
the methodology that has been adopted. There is furthermore reason to believe that the presence
of material on the surface dating to a specific period reflects use of the site during that period.
Additionally, the surface assemblage seems to reflect, if anything, the minimal occupation of a
site. Therefore, when the surface assemblage is combined with information on site morphology,
site sizes may be reasonably estimated. Based on these considerations, it can be suggested that regional surveys that include off-site field-walking and/or analysis of aerial/satellite imagery have
the highest probability of recording archaeological features. Furthermore, in order to achieve reliable site-size estimates per period, the surface of the site has to have been divided into sub-areas
during surveying to allow recognition of shifting patterns of occupation through time. As regards
the problem of interpreting artefact densities and distributions, it will here be assumed, as is actually done by all surveys analysed here, that the recorded sites are settlements rather than specialpurpose sites, unless there are strong indications to the contrary. 27 This assumption may, and in
fact likely will, introduce a major bias in the present reconstruction, but unless more fieldwork is
undertaken there is no way of correcting it. The present analysis is therefore undertaken in the full
awareness that it can only reconstruct very crude trends that are subject to many biases and that
will likely be corrected as new evidence becomes available.
However, it must also be realized that for some regions, archaeological surveys remain the
single most important source of information and provide a much-needed contextualization of
the few excavations carried out in these regions. Despite its evident limitations, survey evidence
is too valuable to be completely excluded from any regional analysis. Concluding, then, it seems
that combining evidence from surveys and excavations has the greatest potential of highlighting
regional trends through time.
4.3 Problems of phase-based chronology
Studies based on survey data rely on material that has been collected from the surface, rather than
obtained from excavations. For late Neolithic and later sites, pottery fragments are generally the
most abundant material, and therefore the most suitable chronological markers. It is assumed that
material present on the surface of a site in some way reflects the use and occupation of that site.
Based on this assumption, surveyed archaeological sites are dated to phases of which artefacts
are present on the surface of the site.28 Without stratigraphic data obtained from excavation, site
occupation phases cannot be more precisely established than the time-depth of the prevailing archaeological chronology allows. If a given pottery type can for example be dated to within a century, the century becomes the smallest unit of time within which occupation signified by that type
of pottery can be dated if only survey evidence is available.
Pottery, as the most abundant chronological marker in survey collections, is generally dated
through stylistic variation, and this will likely lead to the imposition of phases upon culture change,
or rather its material remains. The delineation of phases initially helps to bring order into the vari27 It is for example possible that smaller sites represented temporary camps or agricultural stations that were not occupied year-round, but as Horne has shown (1993: 47), it may be extremely difficult to distinguish between such stations
and permanent dwellings.
28 In the following discussion, and irrespective of how a chronological unit is called elsewhere, the term phase is used
to designate the smallest chronological unit that can be identified in archaeological material culture development. A
phase is usually distinguished by delineating a particular set of features, attributes, or artefacts appearing together
for a certain amount of time. The term phase is preferred above similar notions such as stratum, or level as these
are stratigraphic units that are strictly speaking unrelated to cultural development and certainly cannot be applied to
material yielded by surface surveys.
48
ation observed in the archaeological record, and allows archaeologists to develop a vocabulary to
discuss this variation. At a later stage, however, the drawbacks of phase-based chronology become
apparent (Plog and Hantman 1990). The first drawback is that a phase-based chronology implies
that culture change is step-wise, rather than continuous. By imposing phases on culture change,
it is assumed that phases were stable periods without variation, interspersed by short periods of
change. Secondly, phases can mask variety, especially when their duration is very long. If a long
period of time is divided in few phases, it becomes difficult to study the correlation between culture change and other diachronic processes, for example climate change (Plog and Hantman 1990:
440–1). Thirdly, pottery types that are in use for an exceptionally long period are often neglected in
the process of phase delineation in favour of types with shorter life-spans that better indicate diachronic changes. This may be true from a perspective of chronology construction, but what these
long-living types signify is exactly the opposite, namely that alongside change, there is considerable
continuity between phases. It is only the archaeologist’s predilection for with change that neglects
this continuity (cf. Porter 2007: 72). Finally, there is the site contemporaneity problem, which will
be discussed in the following section. Plog and Hantman are, however, aware that: ‘Phase based
chronologies are thus often a compromise between what is needed and what is attainable, between
a desire to understand as much of the observed variation as possible and the necessity of conveying information to one’s colleagues’ (1990: 440).
4.4 The contemporaneity problem
Many studies of survey data explicitly or implicitly assume that sites dated to the same chronological phase were contemporaneous and were occupied throughout the entire phase. The assumption
of site contemporaneity may be false when sites were only occupied for a shorter period of time
within a phase, and not during the entire phase. As excavations often show interspersed periods
of occupation and abandonment, it is reasonable to assume that sites were periodically deserted,
regardless of the cause. Especially if a phase lasted for multiple centuries, it is likely that sites were
established or abandoned within that phase, but material culture alone would not be sufficient to
detect this shorter lifespan. If all sites from a particular phase are considered contemporaneous,
there is a risk of counting more sites than actually existed at any point in time during that period,
because some sites existed sequentially rather than synchronically. Consequently, population estimates based on site numbers will be too high, and site catchment and territory analyses will be
flawed. Already as early as 1965, Adams recognized in his study on settlement in the Diyala Basin
the problem of overpopulating the landscape with sites in the case of sequential instead of contemporaneous occupations (1965: 124). This observation is variously labelled the problem of overestimated maps (Ammerman 1981: 77), or the contemporaneity problem (Schacht 1984).
The obvious solution to the mentioned problems is to increase chronological discrimination up
to the point that sites can be assumed to be occupied during an entire phase. Chronological discrimination of phases, however, is an ongoing process that only solves the contemporaneity problem in the long term. In the absence of better chronologies, other methods have been employed
to compensate for the contemporaneity problem.
Plog suggested life-histories for each site (Plog 1974 quoted in Weiss 1977). He divided phases
into smaller time units, and assigned total site size to the middle unit of each phase, and smaller
sizes to the initial and final units of each phase. Total site size for the entire phase was then calculated by adding the site sizes of each time unit together. Thus, sites grew and declined within a
single phase. Weiss (1977) suggested dividing the total amount of occupied hectares in a phase by
the length of the phase in years, as a measure of settlement intensity. However, both approaches
still assume that sites existed during an entire phase, although the approach outlined by Weiss is
likely to deflate actual occupied cumulative site size per phase. Sumner (1990) identified contemporaneous settlement by assuming that settlements occupied during two consecutive phases were
inhabited during and around the transition between these phases, and by assuming that sites occupied during three consecutive periods were occupied during the mid-point of the middle phase.
Chapman (1999: 72) proposed that clusters of sites represent sequential occupation of an area
49
by the same small group, favouring residential mobility of a small group over a large sedentary
population as an explanation for the great number of observed sites during specific phases. This
solution assumes, however, that there was a high degree of mobility, and does not compensate for
the duration of individual phases.
4.4.1 Simulating site contemporaneity
Building on earlier approaches, Dewar (1991) suggested a simulation method for establishing contemporaneous site occupations. This method uses assumptions similar to Sumner’s propositions
outlined above. Following the recognition that the transitions between phases represent key points
in the occupational history of sites, Dewar developed a site typology based on the presence or
absence of archaeological material from consecutive phases. Without stratigraphic evidence, the
relation between archaeological finds from consecutive phases remains unknown. Also, the date
of the find within the phase to which it is dated remains unknown. Thus, a find from a hypothetical phase X could belong to the beginning of phase X, whereas a find from subsequent phase Y
could belong to the end of that phase. The assumption that these finds signify occupation for the
entire lengths of phases X and Y runs counter to the presence of significant occupational gaps at
the site (Fig. 4.1).
If a phase lasted for multiple centuries, it cannot be assumed that occupation lasted for the
entire duration of the phase, and occupation for only part of the phase must be reckoned with. As
the phase already represents the smallest unit of time that can be distinguished in the archaeological record, it is impossible to know for how long a site existed within that phase. As a result, it is
also impossible to know which sites existed simultaneously.
The key to solving this problem is to be found in the transition points between archaeological
phases. Two assumptions are made to solve the contemporaneity problem: (a) if two consecutive
phases are present at a site, it is assumed that the site was occupied at least at the transition point
between these phases (cf. Sumner 1990); and (b) if a site was occupied at the beginning and end of
a phase, it is assumed that the site was also occupied during the rest of that phase. Assumption (a)
is based on the idea that sites with material from two consecutive phases have a higher probability of
being occupied during the transition between these phases, than sites with material from only one
of these phases. This concept of probability is then extended in assumption (b), resulting in a ‘dynamic probability model’ of settlement abandonment and establishment (Wilkinson 2000a: 249).
These assumptions can be represented graphically by including the starting and end points of
any given phase as discrete points in time, t1 and t2 (Fig. 4.2). Sites are then classified according
to whether or not they were occupied at any one of these points. In this way, four different occupation types a–d can be distinguished for hypothetical phase Y. A type a site is occupied at t1
but not at t2. This means that the site already existed in phase X, but was abandoned somewhere
during phase Y. A type b site is occupied at t1 and t2, and by assumption (b) it follows that the site
was occupied during the entire length of phase Y. A type c site was occupied at point t2 but not
at t1 so occupation must have started somewhere in phase Y and continued into the next phase.
Occupation of a type d site is not attested at either point t1 or t2, and establishment and abandonment of type d sites must have taken place entirely within phase Y. Changes in the ratios between
these occupation types can potentially indicate long-term settlement pattern changes. For example,
a high ratio of type d sites may indicate a high settlement flux, whereby settlements are regularly
relocated. Conversely, a large number of type b sites could indicate long-term settlement stability
(Dewar 1991).
phase X
real occupation
gap
phase Y
phase Z
gap
reconstructed occupation
Fig. 4.1: Discrepancy between real periods of occupation at a site for hypothetical phases X, Y, and Z, and the period of occupation as it would be reconstructed from survey finds from that site.
50
phase X
X-Y
phase Y
Y-Z
phase Z
aaa
bbbbbbbbbbbbbbbbbbbb
ccc
ddd
t1
t2
Fig. 4.2: Occupation types a–d for hypothetical phases X, Y, and Z, and transition points t1 and t2. Letters a–c indicate the period
of time during which these site types were certainly occupied. Letter d indicates a floating period of time in phase Y during
which the site was occupied. Dashed lines indicate a period of time during which the site may have been occupied (after Dewar
1991: fig. 3).
This procedure of classifying sites can be repeated for consecutive periods and for multiple
sites, whereby t2 of any given phase equals t1 of the following phase. The resulting classification
of sites according to occupation type can be used to generate equations of site abandonment and
establishment rates. It is assumed that settlement abandonment and establishment are relatively
continuous processes and that every site has an equal probability of being abandoned or established. The site abandonment rate Aocc of a given phase is then calculated as the number of sites
that is abandoned during a phase, that is, all type a and d sites, divided by p, the duration of the
phase in years:
(1) Aocc = ( a + d ) / p
The site establishment rate Eocc is similarly calculated as the number of sites established during a
phase, that is, all type c and type d sites, divided by p:
(2) Eocc = ( c + d ) / p
It is now possible to simulate the average number of simultaneously occupied sites during a
phase. This simulation starts with the number of sites existing at t1, a+b, and simulates for every
year until t2 the probability that a site is abandoned and/or established. The site abandonment
probability equals Aocc, whereas the site establishment probability is equal to Eocc. Because of the
method used to calculate Aocc and Eocc, the number of sites at t2 will invariably equal or approach
b+c. For every year, the number of simultaneously existing sites is noted and these numbers are
used to calculate the average number of simultaneously existing sites during that phase (Dewar
1991). In further calculations, this average number is taken to be representative for point t1.5, that
is, the middle of a phase. The number of simultaneously existing sites is now known for points
t1 and t2, respectively the starting and end points of a phase, and point t1.5, the middle of that
phase.29
4.4.2 Evaluating the site contemporaneity model and its results
The site contemporaneity model has several weaknesses resulting from its mathematical and theoretical assumptions. First, the model cannot deal with multiple sequential occupations during a
single phase, as can be illustrated by the following example. When the data on real occupation from
Fig. 4.1 are entered into the model’s format of type a–d occupations, it would result in the site being type c during phase X, type b during phase Y, and type a during phase Z if there were no other
occupation periods. This example shows that Dewar’s model is unable to detect the occupation gap
in phase X. More generally, this means that the model fails to detect multiple occupations within a
29 The simulations were computed with the SetDyn2 computer program that is a slightly modified version of the program provided in Dewar 1991. The main difference is that the SetDyn2 version uses a Poisson distribution instead of
a simple probability for the calculation of the chance that a site is abandoned or established. The program allows the
simulations to be repeated 1000 times per phase to check the stability of the estimates.
51
single phase (Dewar 1994: 150). Thus, this reconstruction is still only an approximation of the real
occupation, but it better reflects the uncertainties about occupation length than the assumption
that the site existed for the entire length of every phase for which material is evidenced.
Furthermore, failure to identify an occupation hiatus from survey data changes the occupation
type of a site. Take a site for which survey evidence indicates a continuous sequence for two successive phases that, when excavated, shows a destruction layer between these phases. This would
mean that a site that was formerly considered a type c site in the first phase, and a type a site in the
second, is now a type d occupation in both phases. Input of these different occupation types in
the contemporaneity simulation will result in slightly lower estimates of contemporary sites in the
case of two type d sites as compared to a type c and type a site. Also, the abandonment rate in the
first phase will be slightly lower, and the site establishment rate in the second phase will be slightly
higher (Dewar 1991: 610–11). Considering the relatively large number of sites, the effects on the
estimates will be minor.
A third problem involves the assumption that sites with material of two consecutive phases are
occupied during the transition between these phases. This assumption contradicts the intuitive idea
that this transition is a period of change, possibly including, or resulting from, abandonment or the
replacement of one cultural group by another. This idea is, however, not necessarily correct and
derives from a conceptual misunderstanding of what a phase represents. An archaeological phase
represents a particular set of features, attributes, or objects appearing together for a certain period
in time. This combination is subject to change, and once significant change has occurred, it can be
designated a new phase. A transition between two phases is, however, only rarely accompanied by a
change of the total associated inventory or all of its attributes. Only if such strong variation is observed, might the transition between two phases may actually coincide with a possible settlement
abandonment followed by reoccupation by another cultural group. More commonly, however, one
or more types will carry over the first into the second phase. The distinction between phases is
therefore primarily a product of the current archaeological discourse, its history, and its conception of change, and only rarely a true reflection of settlement abandonment and reoccupation of
a site. In other words, there is no a priori reason why a site may not be occupied at the transition
between two phases.
Fourthly, the site contemporaneity model assumes constant rates of site abandonment (Aocc)
and site establishment (Eocc) for each phase. This may not necessarily reflect reality as there may
have been moments of regional destruction, for example resulting from earthquakes or conflict.
In these cases, the reconstructed number of simultaneously occupied settlements would be flawed.
Such regional destructions are often reconstructed using a combination of stratigraphical, historical, cultural, and radiocarbon chronologies. The combination of radiocarbon dates with other
types of evidence may, however, lead to ‘suck in and smear effects’. An abrupt event is smeared
out over time through the inherent imprecision of radiocarbon dates. Precisely dated historical
events, on the other hand, tend to ‘suck in’ more poorly dated evidence, especially if that evidence
is radiocarbon dated. Through this process, false event horizons may be created (Baillie 1998). An
example of such an event horizon is the attribution of destructions at several sites to the Akkadian
kings Sargon and Naram-Sin, recorded in texts, evidenced by excavations and supposedly supported by radiocarbon dates. The use of historical texts only to pinpoint destructions is hampered
by the fact that rulers often claimed more deeds than they actually carried out. Finally, cultural
chronologies are rarely sufficiently detailed to date destructions, even at neighbouring sites, more
precisely than 100 years.
Fifthly, Kintigh (1994) proposed to revise Dewar’s model in order to remove several other
inherent flaws. According to Kintigh, Dewar’s simulation of the number of contemporary sites
does not take into account information from type d sites, and the number of contemporary sites
is calculated independently of phase length. He also showed that the standard deviation resulting
from the simulation may be unrealistically low due to the program’s built-in limit of only one site
abandonment and/or establishment per year. Kintigh’s revisions included a fundamental difference between type b sites on the one hand, and type a, c, and d sites on the other hand. His model
was built around the assumption that type b sites are fundamentally different than the other types
because their occupation extends beyond the duration of a phase. Occupation length of the other
52
sites is essentially the same, only the point in time during which they were occupied varies. Kintigh
therefore offered new calculations for occupation length of these two types. His estimations of the
number of contemporary sites are generally higher than those derived from Dewar’s simulation.
In a response, Dewar (1994) argues that Kintigh’s revisions to the original model derive from a
different understanding of the concepts underlying the model. Instead of arguing that type b sites
on the one hand and type a, c, and d sites on the other are fundamentally different as regards occupation length, Dewar conceives these sites as representing different occupation spans on a continuous distribution. In such a distribution, type b sites represent sites with the longest occupation
spans. Also, Kintigh’s revised formulas for calculating site occupation spans sometimes fail to meet
an important criterion for type b sites, namely that their occupation length must be longer than the
duration of a phase. Finally, whereas Dewar’s simulation provides a standard deviation of the estimated number of contemporary sites, Kintigh’s revised formulas do not. The standard deviation
partly reflects the number of type d sites during a phase and signifies changes in site abandonment
and establishment rates. Summarizing, then, there is no need to apply Kintigh’s revised formulas,
although they do represent a viable alternative to Dewar’s original simulation. Both authors agree,
however, that there is a great need for models of site contemporaneity in the study of survey data
(cf. Wilkinson 2000a: 248–9).30
A final objection to Dewar’s simulation may be that it introduces a number of unverifiable
variables, leading to what could be called a floating reconstruction of contemporary sites. These
variables are the assumption of a steady rate of Aocc and Eocc, as well as the derived probability that
a site is established or abandoned. However, the use of these variables does not lead to a reconstruction that is entirely unsupported by the data. At the transition points between phases, that is,
points t1 and t2, the number of contemporaneous sites is entirely based on the presence or absence
of archaeological material. The number of reconstructed sites at points t1 and t2 is therefore not
probabilistic or floating. For point t1.5, the number of contemporaneous sites is partly based on
the presence of archaeological material, that is, the type b sites, and partly on a simulation using
Aocc and Eocc. A distinction can therefore be made between the non-probabilistic nature of points
t1 and t2 on the one hand, and the partly probabilistic nature of point t1.5 on the other hand. Thus,
the degree to which the reconstruction is floating is limited, and there remain points in the reconstruction that directly reflect the underlying archaeological data.
The results from the simulation represent a probable average number of contemporaneously
occupied sites per archaeological phase. In other words, there is always a chance that the actual
number of occupied sites was either higher or lower. In order to allow a better interpretation of
the simulation results, the potential maximum and minimum number of sites per phase will also
be indicated. The maximum number of occupied sites per phase is considered to be equal to the
maximum number of sites that a survey detected for that phase. The minimum number of occupied sites per phase is considered equal to the number of type b sites per phase, as these sites have
the greatest probability of being continuously occupied. The differences between the simulation
results and the potential range will be presented separately for each survey region (Section 5.3).
4.4.3 Estimating total occupied area
Settlements go through cycles of growth and decline, and different states of abandonment, depending on a number of variables, including environment, demography, and social and cultural factors. It is therefore to be expected that settlement size will vary through time, and that not all parts
of the settlement will be occupied at the same time. Houses, or parts thereof, can be both permanently and temporarily abandoned (cf. Otto 2006: 258). Pfälzner (2001: 34) notes that household
cycles emerging from, for instance, birth, mortality, and marriages are a strong motivation for the
modification and/or rebuilding of houses, in turn influencing settlement organization as a whole.
These cycles, however, appear to be relatively short in duration, perhaps lasting no more than one
30 Previous applications of Dewar’s model to survey data from various parts of the Near East can be found in Neely and
Wright 1994, Kouchoukos 1998, Pollock 1999, Lyon 2000, and Ristvet 2005.
53
generation. Environmental cycles may of course last longer (Horne 1994: 79). Horne (1994: 79)
suggests that especially small settlements are likely to suffer from such cycles, due to the relatively
severe impact of demographic changes, as compared to settlements with larger populations.
Settlement location and expansion can be limited by the availability of land for building houses.
If a settlement is located on a tell, the slopes of the tell can limit expansion of the settlement. City
walls can have a similar effect. A study of the village of Qdeir, Syria, indicated that expansion of
this village was impeded by the presence of a cemetery, good quality arable land, and the necessity to reserve open land for the pasturage of animal herds (Aurenche and Desfarges 1983). This
suggests that, when good soils are scarce, and therefore in high demand, they will not always be
used for settlement. Rather, such areas may remain primarily reserved for agricultural or pastoral
production. Settlement will then either be moved to less fertile soils, or to non-arable lands. Also,
when arable land is in high demand, land prices or their equivalents may be so high as to restrict
house sizes and settlement sizes in general (Kramer 1982: 164). Such an environment would result
in highly compacted settlements. Within such an environment, occupants of a site may consider
reusing old parts of a settlement as a viable option when the need arises, for example as a result of
population growth, instead of expanding the settlement onto useful arable land.
The architectural composition of a settlement is very fluid, as is clearly illustrated by Aurenche
and Desfarges’ case-study of Qdeir (1983). Houses are constructed, demolished, rebuilt, or modified according to the whims and needs of the occupants. These activities change the layout of the
settlement through time, and may also impact on the overall population density within the settlement. An interesting observation is the difference in on-site population densities based on the
duration of occupation of that site. Post-medieval, pre-modern Arabic architecture is essentially
agglutinative in nature, leading to cumulative filling in of open spaces in a settlement, especially
because many post-medieval Arabic settlements lacked a central authority regulating the use of
space. This leads to the old town centres being the most densely built-up parts of settlements, with
the highest population densities. Aurenche and Desfarges observe that expansion into previously
unoccupied areas only occurs when all land within the village is built up. This expansion may, however, at the same time be accompanied by abandonment of older parts of the village, presumably
because the newer parts prove more pleasant to live in (Aurenche and Desfarges 1983: 183–5).
When a site’s total occupied area is estimated from archaeological data, the nature of the site
and the post-depositional processes acting upon the material from which the estimate is derived
must be taken into account. Most sites in the surveys analysed here are tells, meaning that older
periods may be obscured by later depositions. This means that parts of or the entire period may
be absent from a surface collection, which in turn influences the estimate made for the occupied
area during that period. There is, however, reason to believe that this problem is less applicable
to the north Mesopotamian Bronze Age than to other periods. Bronze Age occupations at many
tells seem to have been by far the most extensive of all periods, reducing the chance that they were
completely obscured by later occupations. Estimations of Bronze Age occupation sizes seem,
then, more reliable than for many other periods (see Section 4.2).
The estimation of individual site sizes is problematic when Dewar’s method for establishing
site contemporaneity is used. This method establishes the number of contemporaneously occupied sites for three points in every phase, t1, t1.5, and t2. However, point t2 of any phase is equivalent
with point t1 of the following phase. Because site size estimates vary per phase, it is possible that
there are two different estimates for every transition point between phases. There are no objective
criteria to select any of these estimates. Furthermore, the number of simultaneous sites at point t1.5
is an average, and it is not known which individual sites were occupied at this point, except the type
b sites. Thus, it is not possible to calculate the total area that was occupied at this point in time. In a
study on late fourth to early third millennium settlement in Mesopotamia, Kouchoukos (1998) proposed several additions to Dewar’s model in order to solve these problems. Kouchoukos’ model
will be discussed below.
The calculation of total occupied area of all sites within a survey region at point t1.5 is hampered
by the fact that not all sites that were occupied at that point in time are known. This particular
problem of the simulation model can be addressed by dividing the settlement record in size classes
and assigning every site per phase to one of these classes. The site contemporaneity simulation is
54
then performed independently for each size class. The result is an estimate of the number of contemporary sites at points t1, t1.5, and t2 per size class per phase. By assigning an average size to each
size class, it is then possible to calculate the amount of hectares that was simultaneously occupied
at points t1, t1.5, and t2 of every phase and for every class. Finally, by adding up the size estimates
for every class, one arrives at the total simultaneously occupied area for the entire survey region.
In order to come to meaningful size-classes, Kouchoukos analysed long-term datasets from the
Nippur Warka Survey, the Susiana-Deh Luran Survey, and the North Jazira Survey to see if there
were regularities in site sizes. He suggested that sites in these three surveys tended to cluster in
ranges of 1–3 ha, 3–8 ha, 8–15 ha, 15–40 ha, and 40+ ha. Given the fact that c. 80 % of the sites in
the present analysis fall within Kouchoukos’ 1–3 ha size class, new, more fine-grained size classes
will be determined from the surveys used here (see Section 5.3).
The second problem involves differences in site size estimates for any two subsequent phases.
Because point t2 of the first phase is equivalent to point t1 of the following phase, two different
estimates exist for the same point in time. This problem can be avoided by calculating the average
of these two estimates at this point, and using this average in the calculation of total occupied area
per size class. It is argued that this average represents a gradual growth or decline of the site, and
does not necessarily lead to a loss of information.
4.5 Reconstructing population levels
The estimation of population sizes on the basis of Near Eastern archaeological data is notoriously difficult (van de Mieroop 1997: 95–7). There are several methods to reconstruct population
sizes, based on a variety of archaeological, historical, and/or ethnographical data, including artefact densities through space and time, regional carrying capacities, burial ground sizes, population censuses or tax records, settlement sizes, and house counts. Preferably, multiple sources are
used to corroborate the respective results of each individual source (Hassan 1981; Bintliff 1997;
Chamberlain 2006).
It has become increasingly clear that the number of sites per period alone is not an accurate
proxy for the total population during that period, as a few large sites may have a considerably larger
aggregate population than numerous small hamlets (Osborne 2004: 170). In the present analysis,
data on aggregate settlement size will be used to reconstruct population levels. Settlement size as
determined by the occurrence of diagnostic sherds on the surface of the site will be the determining variable of total population. It will be assumed that there were no significant changes in settlement sizes within a period. As suggested earlier, it will furthermore be assumed that all archaeological material in the survey reflects human occupation, rather than specialized activities involving
a temporary stay at a certain location. Where there is doubt about as to the function or permanent
occupation of a site, it is removed from the investigated sample, as is for example the case with a
cultic centre in the Birecik-Euphrates Dam Survey region, and several supposed Early Bronze Age
burial grounds recorded by the Balikh Survey (see Sections 5.2.1, 5.2.2).
Settlement size can only be used as a proxy for population size if both variables are positively
correlated in some way. Various researchers have explored this correlation. Sumner analysed a
sample of 110 Iranian villages and found a good correlation between settlement size and number
of occupants (Sumner 1979: 165). De Roche (1983) studied villages of Mexican subsistence farmers and found the assumption that settlement size correlates positively to population size to be
true as well. Kramer’s analysis of Iranian villages and cities also found a particular strong correlation (Kramer 1982: 158, 163), and an analysis of another dataset provided by Aurenche (1981)
confirms this result. All these calculations suggest that population size and settlement size are to
some degree positively correlated, although there must be additional variables that determine settlement size. Both Sumner (1979: 165) and De Roche (1983: 190) suggested on the basis of their
calculations that the correlation between settlement size and population size is not strong enough
to be applied to individual settlements, but that it could safely be used on a regional level. There
seems therefore to be a basis to use regional aggregate site sizes as a crude proxy for the settled
population living in that area.
55
Whereas Sumner did not list the individual settlements that he used for his analysis, Kramer
(Kramer 1982: tables 5.3, 5.6) and Aurenche (1981: table 3) did provide information on size and
population of individual settlements, allowing the exploration of the relationship between settlement size and population density. The correlation between these two variables is significant at
the 1 % level, but nevertheless weak (rs=0.29), as has also been observed by Kramer (1982: 178).
In fact, variability in population densities is slightly larger in towns and cities than in villages. 31
Thus, although there is a statistically significant correlation between settlement size and population
density, population is obviously not the only variable influencing settlement size. Other variables
may include the need for defence, or the need for space to carry out various activities within the
settlement.
The next step involves the calculation of a population density that can be applied to the aggregate settlement areas for every survey region during each phase. Usually, researchers apply a range
of 100–200 persons/ha to the total settlement area, based on ethnohistorical parallels (Kramer
1982). For third and second millennium bc northern Mesopotamia, several excavations allow a
reconstruction of on-site settlement densities using data from these excavations, rather than ethnographically derived estimates. Such a reconstruction must take into account the possibility that
parts of a settlement were abandoned and that there were differences in population density due to
social or economic distinctions. Below, several previous studies addressing demography and population density in Mesopotamia will be discussed, thereby highlighting the problems associated with
demographic reconstructions, and the great diversity in their results.
Postgate (1994) used a three-step approach to arrive at a range of possible population densities
for the late Early Dynastic (2600–2400 bc) layers of Abu Salabikh (Iraq): (a) estimate total settlement area, (b) estimate the total residential area of the settlement and the number of houses therein, and (c) estimate the number of occupants per house. He argued that the 10 ha main mound
was primarily shaped by the Early Dynastic city wall, so that the currently visible mound is a good
indicator of the past walled area, even though the earliest Early Dynastic layers themselves were 2
m below the current surface of the plain. The area taken up by the city wall was then subtracted
from the total area of the main mound, resulting in 9.5 ha of urban space. This urban space can be
divided in public space and residential space. Public space consists of streets, alleys, and squares,
whereas residential space consists of private houses, which in turn can be divided in house walls,
courtyards, and roofed space. The subdivision into these categories was made on the basis of a
2500 m2 area excavated in the Main Mound of Abu Salabikh. This subdivision suggested that 10 %
of the total urban space was public space (0.95 ha), and the remaining 90 % was residential space
(8.55 ha).32 For the South Mound, a residential space of 6.75 ha was calculated. A similar analysis
of an Old Babylonian residential area in Ur revealed comparable percentages. Postgate then examined six almost completely excavated houses at Abu Salabikh to determine the dwelling space
within each house, that is, kitchens and living rooms. By assuming an average range of dwelling
space per person (4–7 m2/person), the range of persons per house can be determined. By estimating the number of houses in the residential space of Abu Salabikh, it is possible to calculate a
range of persons per hectare of residential space (248–1205 persons/ha). By applying this number
to the total urban space, Postgate arrived at a possible range of c. 3800–18400 persons for the ED
III occupation of both the Main Mound and the South Mound (1994: table 5).
Stone attempted a correlation between Old Babylonian contracts from Nippur, and architectural modifications of the houses in which the documents were found (1981). She found that the
term used for house in these contracts, e2-du3-a, had the more specific meaning of roofed floor
space, suggesting that the actual houses including courtyards were considerably larger. The picture
emerging from the texts and architectural modifications as revealed by excavation is that house
ownership was very fluid and that houses were modified according to the needs of the owners.
31 For the group of villages, SD=43.8 and V=0.51, whereas for the towns, SD=86.9 and V=0.74. All settlements listed by
Aurenche (1981: table 3) and those in Kramer 1982: table 5.3 are considered villages, whereas those in Kramer 1982:
table 5.6 are considered large settlements.
32 Note, however, that the figures used by Postgate are somewhat problematic. According to Postgate 1994, table 1, 87.5
m2 of the 2500 m2 area is taken up by the city wall, but fig. 7 in the same article, illustrating the excavated area on which
the figures in table 1 are based, shows no evidence for a city wall.
56
Stone also tried to link house types to different family units. She suggested that square houses with
rooms on four sides of a courtyard were used by extended families, whereas linear houses with
rooms on one or two sides of a courtyard were used by nuclear families. Analysis of the number
of living rooms per house suggested that each nuclear family required on average one living room,
two other rooms, and courtyard space. When these data are translated into actual numbers of
occupants per house using ethnographic parallels to determine the size of a nuclear family, they
correspond remarkably well with the number of occupants per house that could be expected on
the basis of the textual evidence alone. The number of occupants per house as reconstructed
from the archaeological/ethnographical data translated into 5.31 m2 of roofed space per person.
Although not part of Stone’s analysis, this figure can be extrapolated to a population density of c.
680 persons/ha using Postgate’s data on the relative proportions of roofed versus other residential
space.33
Pfälzner used a similar method for establishing the population of archaeological sites (2001).
For several sites, he determined the total site size and the area taken up by public buildings such as
temples. He then used Postgate’s rule of thumb to determine the actual area taken up by houses.
He studied numerous houses at several sites in northern Mesopotamia and calculated an average
house size, from which the total number of houses per settlement could be determined. Finally,
this number was multiplied by the average number of persons in a household, which he found to
be 5–6 persons, based on ethnographic parallels. In this way, he calculated the number of occupants of the third millennium bc sites of Selenkahiye, Halāwa A, Chuera, and Taya. The resulting
population densities ranged between 300–500 persons/ha.34
Battini-Villard (1999) analysed the Old Babylonian domestic quarter of area AH at Ur. In an attempt to calculate the number of residents of this area, she reconstructed the average family size in
order to apply this value to the number of houses in area AH. Two texts, from the Old Babylonian
and Middle Assyrian periods respectively, provide information on the number of family members.
The first text comes from Mari during the reign of Zimri-Lim and deals with the deportation of
Hurrian families from several villages that were probably located in the Upper Khabur or TurAbdin region. A total of 469 people making up 96 families were deported from five villages, resulting in an average family size of 4.9 persons. The Middle Assyrian list also documents the deportation of Hurrian families. In total one counts 211 individuals, divided across families comprising
from as few as two to as many as eighteen members. The average family consists of 7.8 persons.
Battini-Villard reanalyzed the houses excavated in area AH and reconstructed a total of 55 houses,
leading to an estimate of 270–430 persons for this area. Area AH measured 6750 m2, resulting in
a population density of 400–635 persons/ha. The area enclosed by the city wall, but excluding the
temple area, is 56 ha in size. Assuming that this area was fully built up, these figures would lead to
a population estimate for Ur of c. 22300–35600 persons (Battini-Villard 1999: 388).
Similarly wide-ranging estimates of the number of persons living in a house were made by
Waetzoldt (1996). Using Ur III lists of house inventories and inheritance documents, he found
that houses ranged between 30 and 200 m2, although most families lived in houses smaller than
100 m2. Even larger houses were also documented, but these were invariably associated with workshops or storerooms. As far as can be reconstructed, and taking into account that information on
the number of children of a married couple or head of family is often lacking, between two and
ten persons lived in a house, sometimes including slaves. In some cases, as many as 40 people may
have lived in a single house.
The reconstructions discussed above all concern relatively large, urban sites. As the survey
record analysed here contains numerous small sites as well, it is necessary to reconstruct densities
that may have pertained to such sites. One of the few small settlements to have been exposed over
a large area is Raqa’i level 3 (2600 bc). This level has been exposed over c. 1400 m2, representing
33 The plans of Nippur published in Stone 1987 seem to support Postgate’s dense urban fabric where roofed space is
40 % of the total house area, and houses make up 90 % of the total urban area.
34 In a study on urbanism at Tell Mozan, Pfälzner and Wissing (2004: 81) assumed a density of 400 persons per hectare,
based on the results of Pfälzner 2001. For Selenkahiye, Meijer (2001: 112) suggested a total population of c. 860–1700
persons, based on a size estimate of 10 ha, the observation that c. 1/7th of the excavated area was public space, resulting in a built-up area of 8.6 ha, and an ethnographically derived population density of 100–200 persons/ha.
57
over 45 % of the hypothesized 0.3 ha rural settlement. This undoubtedly atypical settlement was
centred on the large Rounded Building in the south, from which streets and buildings radiated
outwards to the northwest, north, and northeast. As the southern part of the Round Building
was eroded, the expansion of the settlement in that direction could not be ascertained. Several
districts could be distinguished: a temple area, several storage/industrial areas, and a living quarter (Schwartz and Curvers 1992). Pfälzner (2001: 305, pl. 30) identified fifteen separate houses at
Raqa’i, predominantly with one or two rooms. Under the assumption that each house was inhabited by a nuclear family of 5–7 persons, 75–105 people would have lived in the excavated area, translating into a population density range of 535–750 people per hectare. If this density is extrapolated
to the entire site, then 160–225 people would have lived at Raqa’i. There is, however, considerable
room for debate regarding this figure. Pfälzner included the supposed temple as a house, as well as
several areas that can be clearly identified as industrial areas. Furthermore, Pfälzner’s house 2, consisting of rooms 20 and 83, has a total floor area of less than 10 m2. It is difficult to see how such
a small structure would have housed a nuclear family or similar household unit. Nevertheless, this
house is among the clearest examples of distinct architectural house units. However, even when
these houses are excluded, the reconstructed population density for Raqa’i remains relatively high
and well above densities recorded in ethnohistorical sources.
All approaches to on-site population densities have to determine the amount of non-residential
space, including open space, city walls, and public buildings at a settlement. Here, Postgate’s proportions as calculated from Abu Salabikh are often cited, but there is reason to assume that these
figures are not entirely representative. The ratio of residential to public space of 9:1 is calculated
on the basis of a relatively small area of 2500 m2 at Abu Salabikh that is almost entirely filled with
residential structures. If this ratio is extrapolated to the entire site, this leaves little room for public
buildings such as temples or palaces, as Pfälzner also realized. A further possible distorting factor
is that Abu Salabikh is located outside the study area, and that urban layout may have been different in the Jezirah.
The Late Bronze Age Weststadt of Tall Bazi on the Syrian Upper Euphrates may serve as an
example. The Weststadt at Tall Bazi has been excavated over a larger area than Abu Salabikh, and
the excavation was supplemented by magnetometric inspection. The Weststadt represents a planned
extension of the older town on the main mound. The extension was built on a natural terrace and
bordered on the north side by terrace walls, clearly delimiting the built-up area. The quarter existed
between 1250–1180 bc before it was destroyed by fire. This short life-span of the settlement meant
that large parts of the settlement were used contemporaneously and were relatively undisturbed by
later activities. The settlement was structured by several streets and a central square. The quarters
in between were densely built up, but there is evidence for empty parcels, as well as for houses
that were temporarily or permanently abandoned. Some fifty houses were wholly or partially excavated, and magnetometric investigations indicated that non-excavated parts of the Weststadt had
a similar architectural layout. The entire Weststadt may have had eighty houses on a 3.5 ha area
(Otto 2006). Assuming an average household size of 5–7 persons, and one household per house,
this would have resulted in a population for the Weststadt of 400–560 persons, and a population
density of 115–60 persons/ha. This range is much more in line with ethnohistorical population
densities than the Early and Middle Bronze Age estimates. The very crude estimates for site size
(3.5 ha) and number of houses (80), and the average plot size of 132 m 2 (Otto 2006: fig. 169) allow the calculation of the ratio of residential space to public space, which amounts to 3:7. This
ratio is supported by the plan of the excavated area, which shows wide streets and a large central
square (Otto 2006: fig. 9). The figure of c. 30 % of the total area of the Weststadt being devoted
to residential built-up space is by no means very precise, but it is a considerably lower figure than
Postgate’s estimate of 90 %.
Table 4.1 and Figs. 4.3 and 4.4 summarize the archaeological reconstructions discussed above.
The ethnographic data from Kramer (1982) and Aurenche (1981) have been included as well. The
following observations can be made. First, population densities for Early and Middle Bronze Age
settlements in northern Mesopotamia are structurally higher than those observed in (pre)modern
villages and cities. The method used in these reconstructions, that is calculating site size, counting
the number of houses, and assuming an average household size, is equally capable of producing
58
1000000
100000
number of occupants
Ur
Chuera
Abu Salabikh
Halawa A
Selenkahiye
10000
1000
Tall Bazi (Weststadt)
Raqa'i (level 3)
100
10
1
0.1
1
10
100
settlement size (ha)
1000
10000
Fig. 4.3: Scattergram indicating the correlation between settlement size and number of occupants. Crosses indicate the villages
drawn from Aurenche 1981, table 3, and Kramer 1982, table 5.3. Diamonds indicate the towns drawn from Kramer 1982,
table 5.6. Triangles represent archaeological sites. Where necessary, the range of different site size and population estimates is
indicated. Number of occupants and settlement size are given in logarithmic scale.
1200
population density (persons/ha)
1000
800
600
Raqa'i (level 3)
Abu Salabikh
Ur
400
Selenkahiye
Halawa A
Chuera
200
0
0.1
1
10
100
1000
10000
Fig. 4.4: Scattergram indicating the correlation between settlement size and population density. Crosses indicate the villages
drawn from Aurenche 1981, table 3, and Kramer 1982, table 5.3. Diamonds indicate the towns drawn from Kramer 1982, table
5.6. Triangles represent archaeological sites. Where necessary, the range of possible population densities resulting from different
site size and population estimates is indicated. Settlement size is given in logarithmic scale.
name
period
number of occupants
persons/ha
Abu Salabikh
ED III
17.5
3800–18400
215–1050
Chuera
EJ II–III
65
c. 24600
380
Halawa A
EB IV–MB I
16
c. 6260
390
Nippur
OB
no data
no data
680
Raqa’i (level 3)
EJ II
0.3
160–225
535–750
Selenkahiye
EB IV
15
c. 5820
390
LB II
3.5
400–560
115–60
Ur
OB
56
22300–35600
400–635
Table 4.1: Settlement size (ha), number of occupants per settlement, and population density for selected excavated sites in
greater Mesopotamia dating to the third and second millennium bc.
59
results that do fall within the range of ethnographically recorded densities, as is clear from the
reconstruction of the Weststadt at Tall Bazi. Second, there is no clear difference in population density between sites of different sizes, although this may be partly attributable to the small sample
that was analysed here. Third, it is clear that the results of the population density reconstructions
fluctuate widely, and that they do not easily allow the extrapolation of an acceptable average range.
However, the results also indicate that it is quite possible that ethnohistorical parallels do not apply to third and second millennium bc settlements in northern Mesopotamia. Even if only half of
the area of a settlement would have been occupied at a given time, resulting population densities
would still be in the upper range of the ethnographically recorded densities.
4.6 Rank-size analysis
The reconstruction of ancient populations on the basis of survey data allows the reconstruction
of regional population levels. The data on settlement sizes also allow the exploration of the organization of the settlement system. For this purpose, the reconstructed settlement systems will be
subjected to rank-size analysis. Traditionally, rank-size analysis compares an observed settlement
pattern with an expected pattern. This expected pattern is subject to the rank-size rule, which
holds that when sites are ranked according to size, the size of any site with rank N equals 1/Nth
of the size of the largest site.35 Thus, in a settlement system whose largest site is 30 ha, the nextlargest site would be 15 ha, the third-largest 10 ha, and so forth. When this pattern is plotted on a
log-log rank-size plot, it appears as a straight line from the upper left to the lower right corner of
the plot. This pattern is called the log-normal distribution (Fig. 4.5A). If the curve falls off more
sharply than the log-normal distribution, there are fewer large or more small sites than expected,
resulting in a concave or primate distribution (Fig. 4.5B). If there are more large or fewer small
sites than expected, the resulting distribution is called convex (Fig. 4.5C). Various scholars have
proposed combinations of these distributions (e.g. Falconer and Savage 1995), for example the
primo-convex distribution in which the upper part of the curve is primate, but the lower part conforms to a convex distribution (Fig. 4.5D).
4.6.1 Inspecting rank-size distributions
Rank-size analysis is most profitably carried out when the spatial extension of the settlement system is fully known, and when all sites are known within the region covered by the system. This is
rarely the case for archaeologically observed or reconstructed systems. The failure to recognize the
boundaries of a settlement system or the absence of sites from the system can significantly distort the shape of the size distribution. Archaeologists must therefore be particularly careful when
10
size
C
1
A
D
B
0.1
1
10
rank
100
Fig. 4.5: Rank-size plot showing (A) log-normal, (B) primate, (C) convex, and (D) primo-convex distributions.
35 The actual rank-size rule, formulated by Zipf, is expressed as r∙Pq=K, where r is a site’s rank, P is the site’s size, K is the
size of the largest site, and q is the ratio between the force of unification and force of diversification. Archaeologists
routinely assume that q=1, which actually represents a simplification of the original rule (Savage 1997: 233).
60
applying rank-size analyses to a dataset. The degree to which all sites are known within a given
region is difficult to assess. It appears, however, that this is a relatively minor problem for the settlement systems analysed here. Because the size distribution is primarily shaped by the largest sites,
the overall shape will probably remain intact as long as these large sites are included in the system
that is analysed (Johnson 1981: 176). Although it is certain that the archaeological surveys that are
analysed in the following chapter failed to record some sites, it is likely that the missed sites will be
relatively small. Their absence from the size distribution will therefore have only a minor effect on
the overall shape of the distribution.
Defining the boundaries of a settlement system is more difficult. It is clear that many surveys
in northern Mesopotamia were not or only partially bounded by landscape features such as rivers or mountain ranges, which would have likely confined ancient settlement systems as well. It
is therefore highly likely that the settlement patterns observed in survey regions represent only a
sample of a larger system and that there are no clear criteria to assess this sample in terms of its
place in the overall size distribution of the system. This issue is intricately linked with the actual
surveys and their environmental settings that are analysed here. The issue of defining boundaries
of the settlement system will therefore be dealt with in Section 5.3.2, where the individual ranksize analyses are discussed.
The problems of site size estimates and site contemporaneity apply to rank-size analyses as
well. The issue of reliably estimating site sizes has been extensively dealt with above and will not
be repeated here. Just as overfilled landscapes will inflate population estimates, it is likely that such
landscapes will result in the pooling of settlement systems that existed one after another rather
than at the same point in time. As Johnson has shown (1977: 498), pooling of multiple settlement systems will likely result in a convex system even though the constituent systems originally
had other distributions. It is therefore necessary to include a chronological analysis of the settlement system before it is subjected to rank-size analysis. For that reason, the rank-size analysis will
be applied to the datasets that are corrected for site contemporaneity, rather than to the original
datasets.
The statistical significance of observed deviations from log-normality in rank-size plots is usually tested with the Kolomogorov-Smirnov or K- test. There are, however, several problems with
this test when it is applied directly to site size distributions (Savage 1997: 236): (a) the K- test is
normally applied to continuous distributions, whereas site size distributions are discrete, (b) the Ktest assumes that the expected distribution is defined independently from the investigated dataset,
whereas the expected distribution in rank-size analyses is derived from the dataset itself, and (c) the
value at which the K- test finds a deviation to be statistically significant can be incorrect if some
parameters of the sample are unknown and must be estimated.
In order to address these issues, Falconer and Savage wrote the RankSize simulation program
to investigate the chance that a random sample drawn from a theoretical site universe deviates
more strongly from log-normality than the observed archaeological dataset. The number of sites
in the theoretical site universe is determined from the archaeological dataset by first defining the
site recovery rate, or more generally an estimation of the percentage of sites that is known from
the actual settlement system, and by inferring the actual number of sites that must have existed
in the settlement system. Under the assumption that the largest site is known, the site universe is
then created with a log-normal distribution. It contains the inferred actual number of sites in the
settlement system. From this site universe, the program uses Monte Carlo sampling to create random samples containing as many sites as are present in the archaeological dataset. This process is
repeated a number of times, resulting in a percentage that indicates how many times the random
samples deviate more strongly from log-normality than the archaeological dataset. If this percentage is high, it is likely that any deviation from log-normality in the archaeological dataset is not
statistically significant (Falconer and Savage 1995; Savage 1997).
Drennan and Peterson (2004) recently argued that Monte Carlo sampling is not suited to investigating statistical significance of rank-size distributions. In order to address this issue, they
constructed a theoretical site universe consisting of a spatially bounded region containing a large
number of sites whose size distribution conforms to a Poisson distribution. They then randomly
placed spatial sampling blocks of increasing size in this site universe and analysed the size distri-
61
A 10
B 10
C 10
size
size
size
butions in each of these sampling blocks. Drennan and Peterson found that the size distributions
of the sampling blocks, including a 66 % error range, never coincided with the size distribution of
the site universe as a whole, even though this would be expected under normal statistical assumptions. This discrepancy remained when site samples were randomly selected from the site universe,
instead of generated from a spatial sampling block. Even when some measure of organization was
instilled in the site universe by introducing a hexagonal pattern of first- and second-order large
sites, Monte Carlo sampling was unable to approximate the size distribution of the site universe as
a whole. Drennan and Peterson therefore concluded on the use of Monte Carlo sampling blocks
of different sizes that ‘These are simply different analyses of spatial patterning at different scales;
it makes no sense to try to think of the smaller blocks as useful for characterizing the larger population’ (2004: 539). In other words, the chance that a small sample from a settlement system will
adequately represent the entire system is very low.
Instead of the K- test in combination with Monte Carlo sampling, Drennan and Peterson presented the A value with error range as an alternative method for testing the statistical significance
of deviations from log-normality in observed site size distributions. The A value is created by first
scaling the rank-size plot so that the plot has a square shape and the log-normal line a 45° angle.
In this way, the areas above and below the log-normal line are of equal size and they can be assigned a value of 1.0 and -1.0, respectively. The A value represents the proportion of this area that
is taken up by the area between the log-normal line and the line of the observed size distribution
(Fig. 4.6). In this way, convex size distributions will have a positive A value (Fig. 4.6A) whereas
primate distributions will result in a negative A value (Fig. 4.6B). If the observed site size distribution approaches the log-normal line, A will be close to 0. The maximum value of A is 1.0, when
all sites are of equal size. The minimum value of A can be lower than -1.0 if observed sites are
smaller than the expected smallest settlement. The A value does not, however, give information
on the shape of the size distribution because very different size distributions can produce similar
A values. Furthermore, a size distribution with both convex and primate aspects can produce an
A value close to 0, even though the distribution itself deviates markedly from the log-normal line
(Fig. 4.6C). Overall, then, the A value can give a good indication of how much a size distribution
departs from log-normality, although visual inspection of the size distribution remains necessary.
Even more importantly, because the A value is a coefficient of the area between the log-normal
line and the observed site size distribution, it can be used to compare site size distributions from
different periods and with different numbers and sizes of settlements (Drennan and Peterson
2004: 534–5).
In order to test the statistical significance of the A value, Drennan and Peterson suggested the
use of the bootstrap resampling approach (Drennan and Peterson 2004: 539–41). The great advantage of the bootstrap method over the K- test is that the bootstrap does not require knowledge or
assumptions about any expected distribution in order to test the significance of the observed distribution. The bootstrap method takes numerous samples from the original dataset. Each sample
has as many observations as the original dataset, but duplicates the results of some observations
whereas others are omitted. The distribution of the resulting A values are plotted and readjusted
so that they can be used to express a confidence range within which the A value of the original
1
0.1
1
0.1
1
10
rank
100
1
0.1
1
10
rank
100
1
10
rank
100
Fig. 4.6: A values for (A) convex distribution, (B) primate distribution, and (C) primo-convex distribution. Positive and negative
components of A value have dark and light shading, respectively (based on Drennan and Peterson 2004: fig. 2).
62
dataset will probably have fallen. If this confidence zone is wide, it must be recognized that the
original dataset may only approach reality to a limited degree. If the confidence zone is narrow,
the original dataset is probably a good reflection of reality. Furthermore, by calculating different
error ranges for A values of site size distributions that have to be compared, an error range can be
calculated for which the A values do not overlap anymore, and this range indicates the confidence
with which any observation on change in site size distributions can be made. The comparison of
A values with error ranges of site size distributions of different periods thus allows observations
to be made on the possibility that change actually took place.
4.6.2 Interpreting rank-size distributions
After the statistical significance of the observed distributions has been assessed, the distributions
can be interpreted. Various explanations have been offered for the different types of rank-size
distributions (Savage 1997: table 1). Johnson proposed that different rank-size distributions result
from the degree and mode of system integration (Johnson 1980). System integration is defined as
‘the statistical interdependence of change in the population sizes of the settlements of a system,
such that the probability of change in a given settlement is conditionally related to the probabilities of change in other settlements of the system’ (Johnson 1980: 245). This interdependence can
be expressed both vertically, that is, between sites at different levels within a hierarchical system,
and horizontally between sites that are on an equal functional level. In this sense, convex distributions indicate low overall integration or strong horizontal integration, whereas primate distributions indicate strong vertical integration. In log-normal distributions, both horizontal and vertical
integration are strong. More generally, Johnson argued that low integration implied the existence
of relatively autonomous sites within the system, whereas high integration would indicate strong
interdependence and centralization (1980: 245).
A number of explanations focused on the fact that a settlement system may not have been identified correctly, resulting in a rank-size analysis of only a small part, which in turn leads to flawed
results similar to those generated by Monte Carlo sampling (see Section 4.6.1). Johnson pointed
to the possibility that a primate distribution resulted from a partially identified settlement system,
whereas a convex distribution could be the result of the pooling of multiple systems (1977: 498).
Thus, system boundaries were not correctly identified. This problem is particularly relevant for
archaeologists who often work with data from arbitrarily defined survey regions. Other explanations considered the urban history of a given region as critical factor, with primate distributions
emerging in areas with a short history of urbanization. Still others linked rank-size distributions
with core-periphery models, arguing that different distributions will develop in cores as opposed to
peripheries. Falconer and Savage pointed to the possibility that different systems operated at different organizational levels, for example resulting in a primo-convex distribution where a centralized system is superimposed on a lower central place system (1995). Because several of these explanations contradict each other, it is necessary to combine rank-size analyses with as many other
data as possible, in order to be able to argue why a given interpretation is preferred over another
possibly contradicting explanation.
In the present study, site size is expressed in hectares or absolute population figures, so that a
rank-size distribution is actually a measure for the organization of a population in a given landscape. At a very basic level, a convex size distribution indicates that there are many large settlements compared to the number of small settlements. Convexity also indicates that these larger
sites are of roughly similar size; otherwise the distribution would tend towards log-normality.
Convexity, then, is an expression of population dispersal across the landscape in sites that are of
roughly similar size. A primate distribution indicates that there are one or very few large centres
with a large population, and a great number of smaller settlements. The central issue that has to
be addressed here is why households choose to live in a certain community, and not in another. If
a primate distribution is observed, the advantages of living in a large community clearly outweigh
those connected with living in smaller, dispersed communities. Conversely, in a settlement system
with a convex distribution, there are clear advantages to living dispersed over the landscape.
63
The theoretical understanding of site-size distributions as reflections of system integration,
and the actual observation that rank-size distributions reflect population organization can be combined within the framework of competition and cooperation presented in Chapter 3. If there is
competition between communities, integration between these communities would be low, and a
convex site size distribution may be expected. This understanding of competing communities is
consistent with the observation that convexity often results from the pooling of more than one
settlement systems. In other words, convexity indicates the existence of several settlements or
settlement systems that function more or less independently. Conversely, if there are cooperative
mechanisms in operation in a given region, the site size distribution would eventually tend towards
log-normality. Primate distributions are more difficult to interpret. Such distributions indicate
strong centralization, which can arise from cooperation or can be a reflection of the dominant
position of one site over many others achieved by force. Concluding, then, rank-size analysis can
point to certain processes operating within a given region, but the analysis must be combined with
other data in order to fully understand the data at hand.
4.7 Conclusions
The present chapter started by highlighting the problems associated with the interpretation of surface survey results. In the present context this discussion cannot be more than a brief overview.
The discussion has nevertheless shown that some confidence may be placed in survey results, especially when they are combined with data recovered from stratigraphic excavations. The chapter
continued by pointing out the need for a model that explores the degree to which sites within
a chronological phase were contemporaneous, and has detailed and discussed one such model
that fits the data at hand. This model focuses on the identification of contemporary settlements
through the calculation of average settlement life-spans per chronological period. This model will
be applied to the data presented in Section 5.2.
The next part of this chapter dealt with the reconstruction of on-site population levels and
densities. Based on this discussion, the following conclusions can be drawn: (a) there is considerable variation in modern on-site population densities, (b) there is no reason to assume that such
variation did not exist in the past as well, (c) the extrapolation of ethnohistorical population densities to late third-early second millennium bc northern Mesopotamian settlements must be carefully
considered, and (d) the available evidence suggests that third to early second millennium on-site
population densities in northern Mesopotamia were different to, and more specifically on average higher than, ethnohistorically recorded densities. Nevertheless, a reasonable guess as to actual absolute population numbers would be 150–250 persons/ha, including all built-up and open
space. The evidence discussed here suggests that this density may be considered as representing
the lower end of possible ranges for the late third-early second millennium population densities.
The previous discussion has, however, also highlighted the possibility that on-site densities fluctuated through time, as is made clear by the Tall Bazi Weststadt example. The present discussion
furthermore highlighted the fact that settlement sizes, especially aggregated settlement sizes for
entire regions, seem to be a good indicator of changes in the overall population of a region. This
observation further justifies the use of regional settlement data provided by surveys in the reconstruction of demographic patterns.
Finally, Section 4.6 dealt with rank-size analysis as a method to characterize settlement systems
and their organization. A method for determining the statistical significance of observed rank-size
distributions was presented, and a framework was given within which observed distributions can
be interpreted. This framework does not focus on whether or not an observed rank-size distribution coincides with an expected pattern, but aims to identify diachronic changes in the rank-size
distribution. Additionally, the framework aims to explain the direction of these changes, that is
whether the settlement system develops into a more convex or primate direction.
64
5 Regional settlement trends
5.1 Introduction
The method for the reconstruction of simultaneously occupied sites imposes certain requirements
on the regional data set that is used. These requirements are: (a) a survey strategy that is comparable to the strategies employed by all other surveys that are used, (b) a sound chronology with
fixed, absolute dates for every recognized pottery phase, and (c) a clear indication of site sizes per
phase. After the selection of surveys according to these requirements, each survey is described,
and additions or changes to each dataset are presented (Section 5.2). The results of the site contemporaneity simulation and the demographic and rank-size analyses are presented in Section 5.3.
Finally, Section 5.4 will discuss other surveys in northern Mesopotamia, as well as evidence from
historical records, in order to contextualize the evidence from the selected surveys, and to find
whether or not the reconstructed patterns represent unique, local developments, or may be indicative of larger, regional trends. The relation between the reconstructed settlement patterns, climate
change, and the expected social responses will be discussed in Chapter 8.
5.2 Selecting the surveys
The procedures for simulating site contemporaneity, reconstructing population levels, and performing rank-size analyses all require detailed information that is not available for every survey in
the Jezirah. For most surveys, at least preliminary reports exist, but these only occasionally contain
data on individual site sizes. A selection has to be made based on the availability and quality of
the data. Furthermore, survey strategies have to be similar to some extent to allow data comparison between surveys. There are two qualitatively assessable variables that allow selection of suitable surveys, namely comparability of the data, and suitability for the anticipated reconstruction.
Inspection of multiple survey reports suggests that three surveys in and around the Jezirah provide relatively reliable details on site sizes per period for the third and early second millennium bc;
the Birecik-Euphrates Dam Survey (B-EDS), the Balikh Survey (BS), and the North Jazira Survey
(NJS) (Fig. 5.1, Table 5.1). The BS and NJS used comparable strategies, although details may vary
(Table 5.1). Table 5.1 suggests that the survey strategies of the BS and NJS were most similar, with
the B-EDS differing on several points. The B-EDS, which was part of an archaeological salvage
project, was much coarser in resolution. However, the B-EDS has the advantage that numerous
sites were subsequently excavated, thereby providing a measure of control for the sizes originally
assigned to the sites based on the survey results.
The most important characteristic in this context is that all these surveys applied a controlled
sherd collection programme to each site, whereby sites were divided into distinct sampling sectors
from which material was collected separately. This suggests that site size estimates for each survey
region were made on a similar material base. Where possible, the information from the three original surveys will be complemented with data available from subsequent research.
The relatively precise chronology of the BS was largely based on the excavated stratified sequence from Tell Hammam et-Turkman. Although difficulties with this sequence remain, and
high-precision radiocarbon dates that would allow the relative chronology to be fixed on an absolute time-scale are currently lacking, this chronology may be considered sufficiently sound to
be employed without major adjustments. The other two surveys used much coarser chronologies
(Table 2.4). In the case of the B-EDS, later excavations have led to a considerable refinement, as
well as the filling-in of an apparent chronological gap in the settlement record as reconstructed
by the survey. The results of these excavations have therefore been incorporated in the settlement
65
R
A
L
I
P
M
D
B
Q
N O
S
K
E
H
U
T
F
C
G
V
J
2000+ m
1500-2000 m
1000-1500 m
500-1000 m
0-500 m
0 km
intermittent lake
lake
N
30
60
Fig. 5.1: Map of northern Mesopotamia indicating spatial coverage achieved by recent surveys. (A) Birecik-Euphrates Dam
Survey (B-EDS). (B) Tishrin Dam surveys. (C) Tabqa Dam surveys. (D) Land of Carchemish survey. (E) Westgazira survey.
(F) Balikh-Euphrates uplands survey. (G) Middle Euphrates survey. (H) Balikh Survey (BS) and Wadi Qaramogh survey. (I)
Harran survey. (J) Jebel Bishri survey. (K) Wadi Hammar survey. (L) Upper Khabur Survey. (M) Tell Beydar Survey (TBS). (N)
Brak sustaining area survey. (O) 1988 Tell Brak Survey. (P) Leilan Regional Survey (LRS). (Q) Tell Hamoukar Survey (THS).
(R) Northeast Syria survey. (S) North Jazira Survey (NJS). (T) Lower and middle Khabur survey. (U) Khabur Basin Project. (V)
Wadi Ağiğ survey.
number of
number of sites
seasons
area (km2)
remote sensing
imagery
map
survey
years
main sources
A
B-EDS
1989
1
85 (28)
186
no
Algaze et al. 1991, 1994
H
BS
1983
1
210 (86)
450
yes
Akkermans 1984, 1993; Curvers 1991;
Bartl 1994
S
NJS
1986–1990
4
186 (68)
475
yes
Wilkinson 1990; Wilkinson and Tucker 1995
Table 5.1: Survey strategy details for the BS, B-EDS, and NJS. Number of sites refers to the total number of surveyed sites, with
number of sites with evidence for the period discussed here in parentheses. Letters in the left column refer to Fig. 5.1.
record that is used in the present study. The chronology of the NJS was modified from that used
at Tell al-Hawa, which in turn derived from material excavated in the area of the Saddam Dam
salvage project. This pottery chronology was supplemented through surface associations from
demonstrably single-period sites (Wilkinson and Tucker 1995: 89).
5.2.1 The Birecik-Euphrates Dam Survey
The Birecik-Euphrates Dam region differs from the other survey regions analysed here in its environmental setting. The region is not part of the north Syrian steppe plateau but extends into the
Taurus Mountains via the Euphrates Valley (Fig. 5.1A). Concurrent with the Birecik-Euphrates
Dam Survey, a geomorphological assessment of a small area representative of the whole survey re-
66
Regional settlement trends
gion was carried out (Algaze et al. 1994, 6–8). This assessment illustrated the complex geomorphological history of the area, characterized by several alluviation and erosion phases. Five different
zones can be recognized within the valley. The upland plateaus bordering the valley are 600–750
m above sea-level. Although they contain good quality arable soil, settlement today is marginal
due to the very deep water table. The pediments lack perennial water sources, and only one place
was observed where a high water table allowed the use of wells. This area contained a number of
sites. The high terraces at 395–405 m above sea-level resulted from several Pleistocene alluvial and
colluvial phases. The low terraces are 6–18 m above the low water level of the Euphrates. These
terraces are incised by small tributaries of the Euphrates. The soil is generally suited to cultivation, with good access to water. Typical current land use includes a combination of pistachio
trees and dry and irrigated crops (Algaze et al. 1994: 7). Geomorphological observations around
Şaraga Höyük (B-EDS 73), combined with investigations of archaeological deposits, suggested
that the current Euphrates floodplain was shaped in the third millennium, paralleling observations
at Jerablus Tahtani (Sertok and Kulakoğlu 2001: 476; Sertok et al. 2007).
As part of the Güneydoğu Anadolu Projesi (GAP), several dams have been constructed in
the Euphrates, including the Birecik and Carchemish dams. Archaeological surveys were carried
out to assess the archaeological potential of the areas to be flooded. The Birecik and Carchemish
basins were surveyed in 1989. The survey was limited to a 60 km stretch of the Euphrates Valley,
from Carchemish in the south to the modern town of Halfeti in the north. The surveyed area included the Euphrates Valley and extended into the uplands up to a maximum elevation of 400 m
above sea level, c. 10 m above the projected level of the artificial lake. A total area of c. 186 km2
was investigated. The survey started by locating the principle settlement mounds. Around these
sites, pedestrian survey was carried out along transects to detect dependent settlements. 36 Small
sites were located with the help of local villagers, and by transect-walking areas with environmental conditions similar to areas where sites were already known to have existed. Areas along wadi
courses and around permanent water sources were also transect-walked. This strategy resulted in
the recognition of 85 sites, 28 of which yielded third or early second millennium bc material. Most
sites were visited once to collect diagnostics, that is, rims and decorated body sherds. Depending
on the size, complexity, and morphology of the site, diagnostics were collected in multiple sectors
to detect possible settlement shifts through time. Neither aerial nor satellite imagery seem to have
been used during this survey (Algaze et al. 1991, 1994). Periodization suffered from a lack of good
single-period contexts. The surveyors recognized thirteen periods, ranging from the Pre-Pottery
Neolithic to the thirteenth century ad. Palaeolithic remains were also recovered, but not studied as
part of this survey. A later survey in the same region focused on the Palaeolithic remains (Taşkıran
2002).
Occupied area estimates were derived from maximum site size in the case of single period sites,
and whenever possible from the results of the individual sampling sectors in the case of multiperiod sites. When this proved impossible, an arbitrary size of 0.5 ha was assumed for these sites.
It was suggested that this estimate probably resulted in a significant underestimation of total occupied area, but that all periods would be equally affected, leaving the resulting trends through time
in place. Although this is not necessarily true given the fact that the proportion of single-period
versus multi-period sites may be different for every chronological period, there seems to be no
method to correct this bias (Algaze et al. 1991: n. 6).
The absence of evidence for material dating to the second part of the Middle Bronze Age
suggested to the surveyors that habitation of this region was limited to a few strategically located
sites outside the survey area (Algaze et al. 1994: 17). This settlement gap was excluded from the
chronology, which jumped from period 8, Late EBA/Early MBA, to period 9, LBA. For the purpose of the present analysis, which requires a continuous chronological framework, this gap will
here be called Late MB. Subsequent excavations have shown that the B-EDS area was indeed occupied during this period, and the results of these excavations will be incorporated in the present
analysis.
36 The spacing between transects is not mentioned. See Redman (1982: 377) on the impact of transect spacing on site
detection.
67
The settlement record provided by Algaze et al. (1994) serves as the basis for the data set that
will be used here. Rescue excavations at several sites have, however, necessitated some modifications of the picture as gleaned from the survey. These excavations have especially highlighted the
apparent continuity between Late Uruk and Early EBA at several sites, and the presence of Late
MBA assemblages. As far as can be judged from the excavation reports, the survey seems to have
somewhat underestimated the occupations at individual sites. The individual modifications will be
discussed below; the resultant data set of occupied sites per phase is presented in Table 5.3 and
Figs. 5.2 and 5.3.
The recent re-dating of some of the diagnostic ceramics used by Algaze has led to a reevaluation of the apparent Late MBA settlement gap (Peltenburg 2007a: 17, 2007b: 258). Especially
the use of the ‘fruit stand’ stem as a type-fossil for the Late EBA/Early MBA has led to an overestimation of the number of sites for that period, because ‘fruit stands’ seem to have gone out
of use before the final quarter of the third millennium (Sertok 2007: 242–3). A similar argument
can be made for corrugated Hama goblets. Their use continues into the Late EBA/Early MBA
period, but these goblets were especially dominant before that period. This suggests that sites
where they were found could not only date to the Late EBA/Early MBA, but also to earlier periods
(Peltenburg 2007b: 258–9). The re-dating of these diagnostic types in the B-EDS would affect a
significant number of sites, as shown in Table 5.2.
Due to the uncertainties in the dating of these diachronic types, it is difficult to assign sites with
‘fruit stands’ to either the Early or Middle EBA period, whereas those with Hama goblets could
still be dated to the Late EBA/Early MBA. In order to overcome this issue, Peltenburg suggested
re-dating sites with ‘fruit stand’ to the mid-third millennium bc (2007b: 259). This suggestion
will be followed here for those sites where no excavations have taken place, and where no Hama
goblets were present, as these could indicate Late EBA/Early MBA occupation as well. In other
words, the occupations at Near Zeugma 2 (B-EDS 18), Near Hacınebi 1 (B-EDS 27), Kefri Höyük
(B-EDS 38), Şara Mezar Harabe (B-EDS 60), and Near Akarçay 2 (B-EDS 68) will be treated as
Middle EBA, rather than Late EBA/Early MBA. For the excavated sites, the results from the excavations will also be taken into account, and these are discussed below.
site
name
1
Kalemaydanı Höyük
‘fruit stand’ stem
corrugated Hama goblet
2
Karaçalı Höyük
10
Horum Höyük
x
x
16
Tılmusa Höyük
x
x
18
Near Zeugma 2
x
27
Near Hacınebi 1
x
34
Tılvez Höyük
38
Kefri Höyük
44
Zeytinli Bahçe Höyük
x
49
Mağara Höyük
x
51
Mezraa Höyük
60
Şara Mezar Harabe
65
Şavi Höyük
68
Near Akarçay 2
x
69
Gre Virike
x
72
Akarçay Höyük
82
Tıladir tepe
x
85
Carchemish
x
x
x
x
x
x
x
x
x
x
x
Table 5.2: Occurrence of ‘fruit stand’ stems and corrugated Hama goblets on sites in the B-EDS region (Peltenburg 2007b:
table 1).
68
Horum Höyük (B-EDS 10) has been excavated as part of the international cultural heritage
rescue operation preceding the flooding of the Birecik and Carchemish lakes. The site consists of
a 6.4 ha main mound and a 21 ha lower town. A significant part of the main tell has been eroded
by the Euphrates, probably already at the end of the second millennium bc, and Marro gives an
estimate of 10 ha for the third millennium site (2007b: 384). The lower town seems to have been
occupied only in the Roman period. The excavations were carried out on the southeastern slope
and the top of the main mound, and confirmed the periods detected by the American survey, as
well as adding the Early and Middle EBA and Late MBA periods. Architectural features were found
for the EB I and EB IV/MB I periods, whereas the EB II–III periods were attested by substantial
digging activities. The excavators note that the early EB deposit at Horum Höyük is c. 5 m thick,
similar to that uncovered at Zeytinli Bahçe Höyük (B-EDS 44) (Marro et al. 1997, 1998, 2000;
Marro 2007b). Based on these considerations, the size of Horum Höyük will be conservatively estimated at 8 ha during the Early EBA, Middle EBA, and Late EBA/Early MBA periods. 37
Excavations at 12 m high 1.2 ha Tilbes Höyük (B-EDS 14) confirmed occupation for the EB
I and EB III–IV periods, and suggested a possible EB II and MB I occupation. There seems to
be some scant evidence for MB II activities, but the nature of these activities remains unclear. EB
I buildings were found on the southern slope. There is also evidence for outdoor activity areas,
where metal smelting may have been carried out, and some stone-lined tombshave been found.
Radiocarbon dates confirm a late fourth–early third millennium date. EB II occupation seems to
have been much more restricted than the preceding and following phases, as it was not found in
all trenches. An EB III burned building is interpreted as a building with cultic connotations. This
building had several predecessors on the same spot, but with different internal layouts. Extensive
EB IV construction activities are evidenced in all trenches (Fuensanta et al. 1998, 2001, 2002a).
Based on these data, an Early EBA and Late EBA/Early MBA occupation of 1.0 ha will be assumed, as well as a 0.5 ha Middle EBA and 0.1 ha Late MBA occupation.
The large site of Surtepe Höyük (B-EDS 33) was occupied during the LC 5 period, after which
there is evidence for a conflagration. The site was probably immediately re-occupied during the
EB I period. This is also suggested by the similarities in pottery technology for both periods.
There is evidence for at least two large mudbrick platforms at the site that may have functioned
in both the LC 5 and EB I periods. The excavators estimate the LC 5 occupation at 6 ha and suggest that the site may have extended to 20–50 ha during some periods, judging from surface sherd
densities (Fuensanta et al. 2002a, 2006). The periods that are represented by these sherd scatters
are, however, not mentioned. In the absence of more detailed information on these estimates, it
seems quite possible that these densities represent sherd scatters resulting from off-site activities,
for example manuring, in which case the dimensions of the site itself would have been much more
restricted during these periods. Size estimates for the EB I period range from 3 ha (Rothman and
Fuensanta 2003: table 3) to 12 ha (Fuensanta 2007: 143). Based on the LC 5 estimation, the apparent occupational continuity into the EB I period, and the presence of large mudbrick platforms,
the EB I occupation at Surtepe Höyük will be estimated at 4 ha. For EB IV, an estimate of 12 ha
is given (Fuensanta 2007: 143).
Tilvez Höyük (B-EDS 34) is a 3.5 ha site that is c. 9 m high and measures 1.2 ha at the top. The
earliest material found by the survey dates to the late EB period, but excavations brought to light,
among other things, champagne cups, cyma recta bowls and reserved slip ware, suggesting a transitional Late Chalcolithic/EB I occupation as well. EB III is also present in the mound, but there
seems to have been a hiatus during EB II. EB IV architectural remains consist of stone walls, and
there is some evidence for MB I occupation (Fuensanta et al. 2001, 2002a, 2005). Based on these
results, Fuensanta (2007: 143) estimates the size of this site at 3 ha during EB I, 1 ha during EB
III, and 3 ha during EB IV. These estimates will be retained here.
37 Marro (2007b: 396) stressed the strong continuity observed in funerary and ceramic traditions at Horum Höyük, and
suggested that the EB/MB cultural scheme imposes upon the Birecik region a transition that is in reality difficult to
discern in the archaeological record.
69
As part of the archaeological rescue operations, Zeytinli Bahçe Höyük (B-EDS 44) was resurveyed twice, including an intensive survey using 10x10 m collection units. Subsequent excavation
revealed an extremely rich Early Bronze Age I sequence. The Late Uruk–EBA transition seems to
have been very gradual. There are a number of architectural changes indicating different types of
occupations, but these nonetheless seem to have succeeded each other rapidly, as there is no apparent break in pottery. The EB I sequence is characterized by extreme continuity in architecture
and pottery. Architecture consisted of small, simple dwellings. During the later part of the EB I
period, the architecture is of a larger and planned scale, with large rectangular buildings that are rebuilt several times following the same layout. After this occupation episode, the site is abandoned
until EB IV. During the final part of the Early Bronze Age, as well as during the Middle Bronze
Age, settlement was restricted to the main mound. The EB IV settlement is characterized by irregularly laid out mudbrick structures whose functions change through time. According to the excavators, this area may have been on the outskirts of the settlement, again emphasizing the limited
extension of the settlement in that period. Alluvial deposits were occasionally encountered in the
stratigraphy of the site, suggesting that high Euphrates floods may have been responsible for the
size reduction from EB I to EB IV. On top of this, a late Middle Bronze Age wall was unearthed,
interpreted as a fortification wall. Slightly to the east, at the base of the upper mound, a domestic
activity area was uncovered (Deveci and Mergen 1999; Frangipane and Bucak 2001; Frangipane et
al. 2002, 2004; Balossi et al. 2007). Zeytinli Bahçe Höyük will be considered as a 0.8 ha settlement
during the Early EBA period, and as a 0.2 ha settlement during the Middle EBA, Late EBA/Early
MBA, and Late MBA periods.38
At Yarım Höyük/Yarım Tepe (B-EDS 50), excavations brought to light Late Uruk/LC 5, EB I/
Kurban V, and Hellenistic levels. The Uruk and Kurban V levels were highly disturbed by later digging activities. The shallow Uruk/Kurban V deposits indicated a relatively short occupation span,
perhaps 200–300 years. For both periods, parts of several stone houses were excavated, but the
fact that the Euphrates cut away part of the tell suggested that the settlement may have been larger
than that which still stands today. Botanical and zoological material from the site indicated a subsistence economy including agriculture and animal husbandry geared toward local consumption
(Rothman et al. 1998; Kozbe and Rothman 2005). The rather disturbed deposits at Yarım Höyük,
and the fact that an unknown part of the site has eroded away, make any size estimate problematic.
Based on the site map and the interpretation of the site by the excavators as a small self-sufficient
farming community, Yarım Höyük will be considered as a 0.3 ha Early EBA community.
At Mezraa Höyük (B-EDS 51), a second survey found evidence for the Late Chalcolithic, Early
EBA, and MB II periods, which the survey by Algaze et al. did not detect. Cleaning of a large section in the east slope of the mound resulted in the recognition of eight occupation levels, of which
VI to VIII span the Early to Middle Bronze Age. Level VI is dated to the MB II period, but no
architectural features could be assigned to it. The fact that MB II pottery types occurred on most
parts of the site surface, suggests that there was some activity during this period. Both the survey
results and the evidence from the cleaned section suggest that the EB III/IV period was the most
extensive, whereas the LC and EB I/II periods were more restricted (Ökse and Tekinalp 1999;
Ökse et al. 2001; Yalçıklı and Tekinalp 2002). The occupation of Mezraa Höyük will be estimated
at 0.25 ha during the Early and Middle EBA, at 0.5 ha during the Late EBA/Early MBA, and at
0.1 ha during the Late MBA.
Şavi Höyük (B-EDS 65) consists of a seven separate mounds along the Euphrates, numbered
I–VII from south to north. A narrow valley separates Şavi Höyük I from the other mounds.
Resurvey of Şavi Höyük yielded Halaf sherds, no Uruk sherds, few Early EBA, more Middle and
38 The presence of MB I material at Zeytinli Bahçe Höyük is uncertain at the moment. The clear architectural break
between EB IV and MB II suggests either complete abandonment at the site, or the destruction of MB I architectural
levels by the MB II fortifications. Alternatively, the fortification of the site could point to a change in the function of
the site, rather than abandonment followed by reoccupation (Balossi et al. 2007: 376). Radiocarbon dates from in situ
wooden beams from the MB II layers provided an age range of 2028–1980 cal bc. These beams may have been recycled
from older structures that are as yet unidentified at the site (Balossi et al. 2007: 375). Further, the regional MB pottery
assemblage remains problematic. For example, Marro (2007b: n. 25) notes that the pottery assemblage supporting the
MB II date for Zeytinli Bahçe Höyük closely resembles the ceramic repertoire defined as MB I at Horum Höyük.
70
Late EBA, and EB/MB transitional sherds. There were possible LBA sherds as well. Settlement
remains dating to the third and early second millennium bc have only been encountered at Şavi
Höyük I and II, whereas the other mounds seem to have functioned as cemeteries during this period. Excavation at Şavi Höyük I revealed twenty occupation phases, ten of which represent an
almost continuous sequence from late EB to late MB. The earliest phases 20–19 (EB III/IV) consisted of stone foundations, followed by simple domestic structures during phases 18–13 (EB III/
IV–MB I). No architecture was recovered from phase 12, but the large amount of pottery suggests
at least continuous use of the site. Phase 11 (MB II) consisted of simple architecture, whereas
from phase 10 onwards there was again evidence for stone foundations. At Şavi Höyük II, a Late
Uruk building was identified, on top of which Early EBA walls were found that probably belonged
to several buildings. (Dittmann et al. 2001, 2002; Bucak et al. 2004). Based on these considerations, a 0.4 ha size will be assumed for the Middle EBA, Late EBA/Early MBA, and Late MBA
occupations at Şavi Höyük I, and a 0.1 ha size for the Early EBA occupation at Şavi Höyük II.
Excavations at Gre Virike (B-EDS 69) confirmed Algaze’s suggestion (1994: 54) that this was
a special-purpose site. The excavations also confirmed the EB date of the site, but failed to detect
any signs of permanent occupation. Instead, the site seems to have had a purely cultic function
during much of the third millennium bc. The site consisted of a large mudbrick platform with several smaller structures. During period I (EB I–II), there were several plastered and stone-lined pits,
a stone-lined channel, and a stone stairway leading away from the terrace and into a subterranean
stone-lined chamber.39 The stone-lined pits contained grains, ashes, unbaked clay figurines, and
miniature stone axes. During period IIA (EB III–IV), several chamber tombs were constructed on
the platform. Grave goods included numerous vessels, and stone and metal objects. During period
IIB (EB IV), the platform was used for child burials. Associated finds included bronze pins and
numerous vessels. There is also some evidence for outdoor activities on the mudbrick platform.
Gre Virike is interpreted by the excavator as a cultic centre serving the small EB settlements in the
Euphrates basin. There is a change of focus between periods I and II, but the continuity between
IIA and IIB is clear, and strengthened by the re-use of some IIA tombs in the following IIB period
(Ökse 2005, 2006, 2007). This evidence suggests that Gre Virike never saw permanent occupation,
and it will not be used in the present analysis of population dynamics in the Euphrates basin. 40
The surface of Şaraga Höyük (B-EDS 73) has been re-investigated with the aim of excavating
the site. Virtually all the pottery that was recovered from the surface of the tell dated to (post-)
Roman periods, except in exposed sections of the site. These sections showed the presence of
Bronze Age buildings. Early Bronze Age material dated to the mid to late third millennium bc.41
The transition to the Middle Bronze Age seems to have been gradual at the site. 42 During the
Middle Bronze Age, a large building existed with rooms that are interpreted as storage rooms,
based on their inventory of large storage jars. The building was renovated several times, demonstrating a long period of use. Adjacent structures were also well-built, and included a stone stairway. The pottery assemblage yielded a large range of MB types, including types of the later MBA.
Excavation in the western part of the site revealed what the excavators believe to be a fortification
39 The stone stairway is interpreted as leading to a now defunct underground spring (Ökse 2006). Note also the parallels
with the later underground structure unearthed at Tell Mozan and identified as a gate to the netherworld (KellyBuccellati 2002).
40 Gre Virike is not the only non-habitation site in the B-EDS area. Early Bronze Age cemeteries were found near Birecik
(Sertok and Ergeç 1999a), and possibly at the site of Harabebezikan Höyük (B-EDS 75) (Bilgen 2001). Boztarla Tarlası
#2 (B-EDS 24) may also represent an EBA tomb (Algaze et al. 1994: 37). These cemeteries, as well as Gre Virike,
testify to the lively regional social and religious traditions that must have existed in the Birecik area during the Early
Bronze Age.
41 Based on the presence of a nearby EBA cemetery, the excavators also inferred a notable Early EBA occupation at
Şaraga Höyük (B-EDS 73) (Sertok et al. 2007: 343).
42 In two areas of Şaraga Höyük, the excavators noted that the Early and Middle Bronze Age layers were separated by
thick Euphrates flood deposits. Historical records of Euphrates floods document that, once the snow melted, the
water level could rise to as much as 3.5 m above the normal water level in the Birecik area. Thus, before the construction of dams controlling the flow regime of the Euphrates, significant floods may have been the norm rather than
the exception. In fact, EBA/MBA flood deposits were not encountered in the eastern part of the mound, and the
event responsible for them apparently did not completely disrupt life at Şaraga Höyük (B-EDS 73) (Sertok et al. 2007:
348).
71
Early EBA
Middle EBA
Late EBA/
Early MBA
site
name
1
Kalemaydanı Höyük
0.5
2
Karaçalı Höyük
0.25
10
Horum Höyük
8
8
8
14
Tilbes Höyük
1
0.5
1
16
Tılmusa Höyük
18
Near Zeugma 2
21
Tepecik
27
Near Hacınebi 1
31
Tılöbür Höyük
33
Surtepe Höyük
4
34
Tılvez Höyük
3
1
38
Kefri Höyük
0.5
0.5
40
Büyük Tılmiyan Höyük
0.5
0.5
44
Zeytinli Bahçe Höyük
0.8
0.2
49
Mağara Höyük
0.32
0.32
50
Yarım Höyük/Tepe
0.3
51
Mezraa Höyük
0.25
55
Danaoğlu H öyük
56
Elifoğlu Höyük
0.5
60
Şara Mezar Harabe
0.5
63
Cısırın Höyük
65
Şavi Höyük I
(65a)
Şavi Höyük II
68
Near Akarçay 2
0.4
71
Şaraga Höyük
0.5
72
Akarçay Höyük
82
Tıladir Tepe
85
Carchemish
Late MB
0.1
0.5
0.5
0.5
0.5
0.5
1.44
12
0.25
3
0.5
0.2
0.1
0.5
0.5
0.5
0.4
0.4
0.4
0.1
0.5
0.5
12.2
4
30
30
30
Table 5.3: Size estimates (ha) for B-EDS sites with an EB–MB component (based on Algaze et al. 1994, with modifications).
Note that no site number was assigned to Şavi Höyük II by the B-EDS.
wall, probably dating to the early part of the MBA (Sertok and Ergeç 1999b; Sertok and Kulakoğlu
2001, 2002; Sertok et al. 2004). The size of Şaraga Höyük will be estimated at 0.5 ha during the
Middle EBA and Late EBA/Early MBA periods.
The site of Carchemish (B-EDS 85) consists of three parts: the Acropolis mound (3 ha), the
Inner Town (44.5 ha), and the Outer Town (102 ha) (Algaze et al. 1994: 61). Algaze et al. included
Carchemish at an estimated minimal size of 0.5 ha, but the site was not surveyed due to its location
on the modern Turkish–Syrian border (1994: 61). This estimate was at the lower end of the range
of estimates made by various researchers, summarized by Cooper as falling between 0.5 and 44 ha
(2006: 56). Material from the excavations by Woolley has recently been re-analyzed, resulting in a
new scenario for the growth of Carchemish from a small village to a large urban centre with an
upper and a lower town (Falsone and Sconzo 2007). The evidence from the Acropolis clearly suggests that this part of the site was occupied during the early third millennium bc. If the Acropolis
was also occupied during the second half of that millennium, or the early second millennium, any
remains have been destroyed by later levelling and construction activities. On the other hand, the
72
80
N sites + aggregate site size (ha)
70
60
50
40
30
20
10
0
Early EBA
Middle EBA
Late EBA/Early MBA
Late MBA
Fig. 5.2: Total number of sites (bars) and aggregate site size in hectares (line) for every B-EDS period without correction for
site contemporaneity.
1
N
2
0 km 5
10
10
14
16
18
21
31 33
34
27
38
40
44
49
50
55
51
(65a)
65
56
60
71
68
72
63
82
85
Fig. 5.3: Map of the B-EDS region indicating all sites with an Early EBA, Middle EBA, Late EBA/Early MBA, or Late MBA
occupation (based on Algaze et al. 1994: fig. 2).
Inner Town now provides clear evidence for EB III–IV occupation, especially in the area of the
Great Staircase (Falsone and Sconzo 2007: 88–9). The construction of the wall around the Inner
Town cannot be dated with certainty, but Falsone and Sconzo suggest that this may have occurred
around the same time that the Inner Town became occupied, based on the occurrence of EB
III–IV wares in the fill of the rampart (2007: 90). Peltenburg in turn used this argument to suggest that the greater part of the Inner Town must have been occupied during the final quarter of
the third millennium, since it seemed unlikely that the earth for the fill would have come from a
large distance (Peltenburg forthcoming). The evidence for Late MBA occupation is equally scanty,
with no material of this period attested on the Acropolis, and only some minor occurrences in
the Inner Town (Falsone and Sconzo 2007: table 5.1). During the Mari period, Carchemish seems
to have formed an alliance with several nearby cities, but the city eventually appears to have lost
its independence to Halab (Hawkins 1980). Based on this evidence, the size of Carchemish will
be estimated at 4 ha during the Early EBA, that is, the Acropolis mound only. The Middle EBA,
Late EBA/Early MBA, and Late MBA occupations will be conservatively, and very provisionally,
estimated at 30 ha. This estimate acknowledges both the attested occupation in the area of the
Great Staircase and the uncertainty of its expansion toward the Inner Town rampart, as well as the
continued importance of Carchemish as a political centre.
73
5.2.2 The Balikh survey
The Balikh Valley took shape during the Upper Pleistocene, but the river acquired its present
course only at the end of this period due to tectonic activity (Mulders 1969: 41). Tectonic movement has resulted in a gradual rise of the northern part of the Balikh basin, whereas the south
experienced a gradual sinking (Mulders 1969: 36). On average, the valley is 2–3 km wide, with a
maximum of c. 6 km. Within the Balikh basin, two terraces can be distinguished; a lower Holocene
terrace (BT1) and a remnant higher Upper Pleistocene terrace (BT2). The Holocene terrace consists of brown loam, over 3 m thick with some local marl outcrops (Mulders 1969: 49). Remnants
of the higher Pleistocene terrace are located on the western side of the Balikh Valley and consist
of limestone overlain by gypsum and gravel layers. Limestone outcrops can be found on higher
hilltops (Mulders 1969: 50). The Balikh cuts a floodplain 150 m wide and 4–5 m deep. Both depth
and width of the Balikh floodplain decrease after Tell es-Seman due to tapping water for irrigation
purposes. After the confluence of Balikh and Wadi al-Kheder, the floodplain cuts 3 m deep into
the lowest terrace (Mulders 1969: 8–9). Alluvial sedimentation since the fifth millennium bc has
resulted in a considerable rise of the flood plain and may have obscured low-relief archaeological
features (Wilkinson 1998: 65).43
The Balikh cuts its way through a landscape of undulating plains with predominantly calcareous and gypsiferous soils. To the west of the valley, and in the northern part of the east bank area,
marl deposits predominate, covered by a gravel layer of variable thickness. The southern part of
the east bank area consists of gypsiferous soils, generally covered by a thin layer of gravelly loam
(Boerma 1988: pl. 2). In these areas, low soil aggradation has resulted in high visibility of archaeological features. Visibility decreases towards the Balikh Valley (Wilkinson n.d.). Highly gypsiferous
soils are unsuitable for plant growth, but wadi soils generally have a lower percentage of gypsum,
resulting in local, slightly more favourable circumstances. Calcareous soils are more favourable for
plant growth.
The Balikh is a relatively small perennial stream compared to the Euphrates and Khabur (see
Section 2.2). The Balikh receives most of its water from the spring at ‘Ain al-Arûs, just south of the
present-day Turkish–Syrian border. Additionally, the Balikh is fed by a number of wadis, including the Wadi Qaramogh and Wadi al-Kheder. Most wadis east of the Balikh Valley drain towards
the Balikh and run east–west. In the south, wadis drain towards the Euphrates. These wadis can
transport considerable amounts of water, as evidenced, for example, by large limestone blocks
found in the lower course of the Wadi al-Burj. The Wadi Qaramogh drains a vast area to the west
of the Balikh Valley. The Wadi al-Kheder is fed by the Wadi al-Hemar, coming from the northern
limestone plateau, and the Wadi al-Burj. The depth of the groundwater table varies from 5–10 m
for the lower Balikh Valley to 30 m for the surrounding limestone and gypsum regions (Mulders
1969: 48–9).44
Archaeological interest in the Balikh Valley is a recent phenomenon. Sachau, one of the few
early travellers to have actually visited the Balikh Valley, mentions that there are no ancient monuments in the area (1883: 241). Le Strange mentions some Islamic monuments (1905: 104). Sykes,
travelling in the early twentieth century, mentions ancient irrigation canals between Tell as-Saman
and Raqqa and observes of the Balikh that ‘The numerous ruins along its banks show that a very
large population must at one time have subsisted on both sides of the river’ (1915: 314). Interest in
the archaeology of the Balikh Valley started with Mallowan’s soundings at Aswad, Hammam IbnShehab, Jidle, Mefesh, and Sahlan (Mallowan 1946). In Mallowan’s words, finds from these sites
‘merely amplify and confirm the evidence already obtained from the Habur’ (1946: 111), and further led him to conclude that ‘a country such as the Balih is likely to have kept within its confines
only the more primitive and least enterprising forms of society’ (1946: 115). Compared to the scale
of settlement in the Khabur area, the Bronze Age sites of the Balikh Valley are indeed modest. As
43 The effect of this alluviation on Early and Middle Bronze Age settlements is as yet unknown. Given the preferential
location of especially mid–late third millennium bc settlements on tells, a case could be made that the high visibility
of this settlement location at least partially offset the effects of alluviation.
44 Recent canalization of the Balikh and the large-scale use of mechanized pumps have drastically altered the hydrological qualities of the wider Balikh drainage.
74
a result, prehistoric layers at multi-period sites are more exposed, and later surveys of (parts) of
the Balikh Valley showed that the area was home to multiple prehistoric societies (Cauvin 1970;
Abu Assaf 1978/9; Copeland 1979, 1982). A more ambitious survey, targeting the entire Syrian
part of the Balikh Valley, was undertaken in 1983 by a team from the University of Amsterdam
as part of the excavation project of Tell Hammam et-Turkman (Akkermans 1984). Following
this survey, surveys of (parts of) the Balikh Valley, as well as the Harran plain to the north, were
undertaken by Córdoba (1988), Yardimci (1993) and Wilkinson (1998) (see below). Excavated
material for the third and early second millennium bc comes from Tell Bi’a (BS 1) (Einwag 1998;
Miglus and Strommenger 2002 and references therein), Tell Hammam et-Turkman (BS 175) (van
Loon 1988; Meijer 1989, 1996), and from small soundings at Sahlan (BS 247) and Jidle (BS 275)
(Mallowan 1946).45
The 1983 Balikh Survey (BS) (Fig. 5.1H) has been partially published (Akkermans 1984; Curvers
1991; Akkermans 1993; Bartl 1994). The purpose of the survey was to locate as many archaeological sites as possible, and to reconstruct ancient settlement patterns and their development.
The survey also created a regional context for the stratified ceramic sequence of Tell Hammam
et-Turkman. Recognition of sites did not take place in the field, but was based on the screening
of available maps and aerial photographs taken in the 1960s. This resulted in the recognition of
210 archaeological sites, as well as some mounds that were non-archaeological in nature. These
sites were then visited and sherds and other materials were collected in a sampling manner. Large
sites were divided into multiple sampling areas based on morphological characteristics of the site.
No surveying was undertaken away from the sites that were identified prior to the actual field
walking. It is important to note that this technique is biased towards large, recognizable mounds.
Small sites, representing only one period or other limited activities, are likely underrepresented
(Akkermans 1984; Curvers 1991).
Based on the survey results, Curvers (1991: 180) estimated site sizes by dividing each site into
sectors and adding up the sectors that yielded sherds of a specific period. It seems that this division was based on the divisions used during the survey. The survey used a chronology adapted
from the Hammam et-Turkman sequence, in order to avoid any chronological or cultural implications attached to the adoption of a chronology from another region (Curvers 1991: 181). The
chronology used here is modified after the periodization presented by Curvers (Table 2.4).
In 1986, a survey of the Balikh basin was carried out by a Spanish team from the Universidad
Autónoma of Madrid. This survey was part of a project on ‘the origins, development and extinction of the Hurrian culture and people’, and specifically aimed at finding a suitable site for excavation. The survey was restricted to twenty-six tells that were identified on maps and Landsat satellite
images. Many of these sites were previously visited by Akkermans in 1983, but some new sites
BS
name
BS periods
Córdoba periods
2
Córdoba
1
Tell Zaidan
VIab, VIcd
EB IVA
83
7
Tell as-Saman
VIab, VIcd, VIIa, VIIb, VIIc
EB III, EB IVA, EB IVB, MB I
147
11
Tell as-Sawwan
VIab, VIcd, VIIb
EB III, EB IV, MB
152
13
Tell ez-Zkero
VIIa, VIIb, VIIc
EB IV, MB I
175
14
Tell Hammam et-Turkman
EB III, EB IVA, EB IVB, MB I, MB II
247
19
Tell Sahlan
EB III, EB IVA, EB IVB, MB
275
24
Tell Jidle
VIab, VIcd, VIIb, VIIc
EB III, EB IV, MB
276
25
Tell Hammam Ibn-Shehab
VIIb
MB
280
26
Tell Abyad
EB IV, MB
297
23
Tell Breri
VIab, VIcd, VIIb, VIIc
EB IVA, EB IVB, MB
Table 5.4: Third and second millennium bc chronological concordance for sites surveyed by the BS and Córdoba. Site numbers
in the second and third columns refer to those published in Curvers 1991 and Córdoba 1988, respectively.
45 Surrounding areas that were recently surveyed include the upland area between the Euphrates and the Balikh (Fig.
5.1E, F) (Einwag 1993; Danti 1997, 2006–7), the Wadi Hammar (Fig. 5.1K) (http://web.uni-frankfurt.de/fb09/
vorderasarch/survey.htm (Accessed 16 November, 2007)), the Euphrates Valley west and east of Raqqa (Fig. 5.1G)
(Kohlmeyer 1984), and the Harran plain (Fig. 5.1I) (Yardimci 1993).
75
located on the Wadi Qaramogh were surveyed as well. No information on on-site collection strategies is provided (Córdoba 1988). Generally, their findings seem to confirm those of Akkermans
for the tells that were visited by both teams, except that the early EB period is conspicuously absent from Córdoba’s data (Table 5.4). Lack of familiarity with local EB pottery traditions, or the
overburden from later periods, particularly Balikh VIcd, possibly combined with a low-intensity
survey strategy may explain this absence. Unfortunately, data on tell morphology and settlement
size per phase are lacking, preventing inclusion of this survey in the present analysis. 46
The Western Jazira Archaeological Landscape Project investigated the Balikh valley from 1992
to 1995, focusing on the northern part of the valley between Tell Sahlan (BS 247) and Tell asSaman (BS 83) (Wilkinson 1998). The project was aimed not only at finding new sites, but especially off-site features relating to irrigation and land-use. Maps and aerial imagery were studied to
identify archaeological and environmental features, focusing especially on those areas and features
that had not been studied by earlier surveys. These features were then visited in the field, together
with features that were discovered during fieldwork. Some intensive fieldwalking around selected
sites was carried out as well. Irrigation features were only sporadically encountered and were generally of Byzantine or later date, if they could be dated at all. Only one feature possibly dating to
the Early Bronze Age was found. Among the other features that were found were hollow ways, and
field scatters of sherds. The hollow ways seemed to date mainly to the Iron Age and early Islamic
period, whereas the field scatters consisted of Hellenistic to early Islamic sherds. Several new,
small sites that were recognized were given BS numbers in accordance with the system used by
Akkermans. The survey found several Halaf and Ubaid sites and to a lesser extent also Bronze Age
sites. Only three new rural EB/MB sites were found, two of which may have been special purpose
sites not associated with agricultural activities or permanent occupation. This finding significantly
strengthens the picture of third and early second millennium occupation in the Balikh Valley as
gained from the 1983 Balikh Survey (Wilkinson 1998: 71).
Lyon visited numerous sites in the Balikh Valley as part of the Western Jazira Archaeological
Landscape Project, and presented data on site sizes for periods VII–IX that are sometimes at odds
with Curvers’ estimates (Lyon 2000: table 2). In some instances it can be clearly seen that Lyon
used Curvers’ estimates, in which case they are in accordance. At several sites, however, there are
considerable differences between Curvers’ and Lyon’s estimates for period VII occupations (Table
5.5).
site
name
69
no occupation
73
83
Tell as-Saman
84
100
Tell Khadriya
estimate Lyon (ha)
2.1
0.6
1.3
9.0
17.0
10.0
Tell Jaruhe
112
139
estimate Curvers (ha)
no occupation
0.2–1.0
2.2
0.6
2.5
0.5
6.3
182
0.2
205
1.8
1.3
no occupation
219
Tell Qardana
0.6
1.8
247
Tell Sahlan
7.5
16.0
316
Tell Mafraq Sluk
0.3
1.6
0.6
1.2
334
Table 5.5: Comparison of widely varying site size estimates by Curvers and Lyon for Balikh period VII.
46 Work in the Wadi Qaramogh area has recently been resumed with the archaeological investigation of Tell Balabra East
and West/Tell Barābira 1 and 2 (http://www.jezireh.org/balabra_excavation.html (Accessed 5 March, 2009)).
76
Furthermore, Lyon mentioned two Balikh period VII sites that were not found by the 1983
survey. These are BS 393 and JDL-1, estimated at 1.5 and 1.8 ha, respectively. In the absence of
information on the assumptions according to which site sizes were estimated by Lyon, it is difficult to accord Curvers’ and Lyon’s estimates with each other. It seems that Lyon’s estimates are
consistently higher than those of Curvers, although there are some notable exceptions. A possible
explanation may be that Lyon’s higher estimates resulted from the lumping together of period VII
material, which Curvers subdivided into three phases. Given these uncertainties, it seems most
prudent to use Curvers’ data, as these data are at least internally consistent.
Before site contemporaneity and population reconstructions could be carried out, a site gazetteer was compiled from Curvers 1991, listing all sites with an Early and/or Middle Bronze Age
component, that is, Balikh periods VI and VII. This list was then screened for sites that may not
have been in use as habitation sites. These sites were removed from the sample. For some other
sites, revised size estimates are presented. The modifications and additions that were made to the
original data set are described below; the resultant data set is presented in Table 5.6 and Figs. 5.4
and 5.5.47
Tell Bi’a (BS 1), ancient Tuttul, at the confluence of the Euphrates and the Balikh, was not
surveyed in 1983, but an earlier pre-excavation survey indicated that the entire tell was littered
with Ur III and Old Babylonian sherds (Strommenger 1981: 26). An important size indicator is the
city wall, parts of which have been excavated at the northern, western, and southern limits of the
tell (Miglus and Strommenger 2002). At least three phases have been excavated. From the earliest
phase onward, the wall was rebuilt at the same location, possibly so that the older wall could serve
as a foundation for the new structure. The dating of these phases is difficult, as few finds could
be linked to the actual construction of the walls. The earliest material from hill E, datable to the
ED period, is not found in association with the earliest phases of the wall, suggesting that the
first phase dates to the late ED period. Pottery from stratigraphic contexts between the first and
second phase of the wall also dates to the ED period, whereas an Akkad-period grave dug into
the second phase provides a terminus ante quem. The available evidence therefore suggests that the
oldest and middle phases were constructed late in the ED period. The third phase seems to date
to the late Isin-Larsa/early Old Babylonian period. If it is assumed that the city wall partly shaped
the contours of the tell, as is suggested by the excavated parts, then the wall would enclose an
area of 30–35 ha. Only a fraction of this area has been excavated, with a focus on public/official
120
100
80
60
40
20
0
Balikh VIcd
Balikh VIab
Balikh VIIa
Balikh VIIb
Balikh VIIc
Fig. 5.4: Total number of sites (bars) and aggregate site size in hectares (line) for every BS period without correction for site
contemporaneity.
47 No attempt has been made here to systematically transcribe Arabic site names according to one system. For wellknown sites, the most widely used transcription is adopted, e.g. Leilan instead of Laylān. Otherwise, the names of
sites, as far as known, are directly taken from the respective publications, to allow easy referencing between these
publications and this study. The APUM project provides a commendable attempt to collect, categorize and systemize
site names in northern Mesopotamia (Anastasio et al. 2004: 2–3).
77
site
name
VIab
VIcd
VIIa
VIIb
VIIc
1
Tell Bi’a
20
20
15
15
15
2
Tell Zaidan
7
0.5
12
21
Tell Mahlas
2
3.2
27
1
1
34
35
Merj Abu Sharib
56
0.2
0.8
0.8
66
0.2
1.5
1.5
67
0.5
0.5
0.5
73
0.6
0.6
0.6
81
Khirbet al-Bassal
0.5
1.2
1.2
83
Tell as-Saman
2.7
4.5
9
9
9.7
84
10
10
10
10
85
0.4
86
0.3
95
Tell ar-Rahyar
0.5
0.5
96
0.1
0.1
100
Tell Jaruhe
0.8
1
101
Helwet’Abid
105
1
106
2.1
1.5
112
0.6
0.6
113
139
Tell Khadriya
1.4
0.4
0.5
147
Tell as-Sawwan
2
2
0.3
152
Tell ez-Zkero
171
Khirbet al-Hajaje
0.4
172
Khirbet el-Ambar
1
175
Tell Hammam et-Turkman
3
3.7
180
1.5
1.5
181
1
1
182
186
195
200
0.1
0.8
1.3
1.1
1.1
1.1
0.1
0.4
0.3
0.3
0.3
0.2
1
1
1.4
0.6
1.1
0.5
0.5
0.5
0.5
6.6
6.6
6.6
0.2
0.2
0.2
0.2
2
2
2
0.3
0.3
0.3
203
0.7
0.7
0.7
205
1.8
1.8
1.8
207
0.2
0.2
0.2
211
Tell Jital
4.2
4.2
4.2
212
Tell Eftaim
3
1.7
214
1.3
3.5
Table 5.6: Size estimates (ha) for BS sites with a third millennium
VIIa–c) component (based on Curvers 1991, with modifications).
78
0.2
bc
(Balikh VIa–d) or early second millennium
bc
(Balikh
site
name
VIab
VIcd
VIIa
VIIb
VIIc
219
Tell Qardana
0.6
0.6
235
Tell Gharair/Tell Wazgöl
0.8
236
1
240
241
242
Tell ed-Dubaghiye
244
246
1.6
3
1.4
3
1
247
Tell Sahlan
2.3
2.3
7.5
7.5
7.5
254
264
Tell Shrey’an
1
265
0.2
267
0.3
268
0.4
270
Tell al-Abed
0.7
0.3
275
Tell Jidle
1.5
1.5
276
Tell Hammam Ibn-Shehab
280
Tell Abyad
281
282
284
Tell Ksas
1
1
297
Tell Breri
3.8
3.8
304
Maslahat Tell as-Saman
305
Khirbet al-Halawa
308
Tell Sabi Abyad V
310
0.1
0.2
0.5
0.3
0.6
0.2
2.4
2
2
0.2
0.7
0.7
0.7
1.5
1.5
0.9
3
4.8
4.8
4.8
4.8
0.2
0.6
0.6
0.6
3.8
3.8
0.3
0.7
0.7
0.2
0.2
0.2
0.5
0.5
311
0.1
316
Tell Mafraq Sluk
0.3
317
0.2
322
323
324
326
327
Tell Abbara
328
330
331
334
337
340
0.5
342
1.3
0.3
0.2
0.2
0.3
0.3
0.3
1.1
1
1
1
1
1
1
1
0.1
0.1
1.4
1.4
1.4
1.4
0.6
0.6
1.6
1.6
0.3
1.4
0.6
0.6
1.6
Table 5.6: Continued.
79
N
330
280
282
0 km 5
281
10
331
276
297
275
270
284
264
326
265
323
268
324
267
246
241
242
240
214
236
334
337
340
182
183
181
180
247
244
254
235
211
212
322
200
195
219
207
186
308
175A
175 171
205
172
305
203
152
310
316
311
147
139
317
112
342
86
304
67
73
113
106
85
105
84
83
81
66
328
327
56
35
34
101
100
96
95
27
329
21
9
1
5
2
12
3
Fig. 5.5: Map of the BS region indicating all sites with a Balikh VIab, VIcd, VIIa, VIIb, or VIIc occupation.
buildings, complicating the estimation of built-up area within the city wall. Based on the dating
of the various palaces and temples, it seems reasonable to assume that Tell Bi’a was largest during
the mid-third millennium bc, that is, late Balikh VIab. Comparison of Tell Bi’a with urban sites
upstream of the Euphrates, notably in the Tabqa/Tishreen Dam areas (Cooper 2006: table 3.1),
suggests that the maximum size of Tell Bi’a may have been 15–30 ha. Therefore, a conservative
maximum size of 20 ha is estimated for the Balikh VIab and VIcd periods. The Balikh VIIa–c oc-
80
cupation is estimated at 15 ha, reflecting both the repeated destructions of palace A by YahdunLim and Zimri-Lim, and the continuing importance of Tell Bi’a as a regional capital during the
early second millennium.
Tell Zaidan (BS 2) was surveyed by Kohlmeyer (1984: 108), the 1983 Balikh Survey (Curvers
1991: 183), and Córdoba (1988: 156). The major occupation of this site probably fell in the Halaf
and Ubaid periods. Whereas Kohlmeyer and Córdoba suggested a possible early EB occupation,
Curvers assumed full Balikh VIab occupation and a minor Balikh VIcd occupation. A possible
early EB burial ground has been identified near the site (Kohlmeyer 1984: 108). Copeland (1979:
fig. 15) reported a possible Khabur ware sherd, but as none of the other surveyors found this pottery, it may have been either a stray object or a misidentification. Based on the drawing published
by Kohlmeyer (1984: fig. 4), Zaidan can be estimated at c. 10 ha, which will here be translated in
size estimates of 7 and 0.5 ha for the Balikh VIab and Balikh VIcd periods, respectively.
Strommenger (1981: 34) mentioned an unnamed site northwest of Tell Bi’a, where Old
Babylonian sherds were found. However, the site is not indicated on the map of the surroundings
of Tell Bi’a published by Schirmer (1987: figs. 1, 3). Curvers mentioned this site as BS xx, but it
was apparently not visited by the 1983 Balikh Survey (1991: 183). As no location or any other details are known for this site, it will not be included in this study.
The BS recorded several sites that are interpreted as burial grounds rather than settlements on
the basis of the presence of disturbed graves and their location on the edge of the valley. All supposed burial grounds date to the Balikh VI period, that is, the third millennium bc, and include BS
3, BS 5, BS 9, BS 105, BS 106, BS 112, BS 175A, and BS 329. Extramural burial grounds for this
period are known from the Upper Syrian Euphrates area (Akkermans and Schwartz 2003: 251–3).
Meijer (2007: 319) noted the conspicuous absence of EB burials at Hammam et-Turkman, suggesting that the dead may have been buried outside the settlement. It has been proposed that under
certain circumstances extramural burial grounds may be used as a substitute for otherwise invisible
settlements (Renfrew 1972: but see Cherry 1979). The vicinity of the Balikh Valley burial grounds
to known contemporary settlements suggests, however, that it is unnecessary to add sites to the
data set on the basis of the reported burial grounds.48
The archaeological nature of some sites identified by the BS was questioned by the Western
Jazira Archaeological Landscape Project. BS 221 consisted of pale grayish brown fine sandy silts
with very few sherds, suggesting that this site represented the result of digging activities rather
than a true habitation site. The same was suggested for BS 243 where virtually no pottery was
found (Wilkinson n.d.). Therefore, these two sites were removed from the sample.
Further modifications of the original Bronze Age data set include the removal from the sample
of the Balikh VII sites Khirbet esh-Shenef (BS 170) and Tell Sabi Abyad I and III (BS 189 and BS
191, respectively), as excavations suggested that Balikh VII was not present at these sites, although
Balikh VIII was (Akkermans 1993).
5.2.3 The North Jazira Survey
The area consists of east–west orientated anticlinal ridges, of which the Jebel Sinjar is the largest.
The entire area slopes, resulting in runoff from rainfall and subsequent wadi formation once the
soil is saturated. These wadis are deeply incised and severely restrict movement, as does the moderately incised Tigris River to the north and east of the study area. To the north, the area becomes
hillier towards the Taurus/Zagros Mountains. Modern climate is a variant of the warm temperate
semi-arid climate. The long-term mean annual rainfall for the north Jezirah, based on records from
Mosul, is estimated at 370 mm. On average, there is a drought every ten years, resulting in less than
250 mm annual rainfall. The distribution of rainfall across the year, however, allows farmers to
48 BS 3, 5 and 9 may be associated with either Tell Zaidan (BS 2) or Tell Bi’a (BS 1), BS 105 and 106 with Tell as-Saman
(BS 83), BS 175A with Hammam et-Turkman (BS 175), and BS 329 with Tell Mahlas (BS 21). Only BS 112 seems to
have no nearby settlement counterpart although the site could possibly belong to Tell as-Saman (BS 83). Alternatively,
the burial grounds may have been used by a (semi-)nomadic population of the region.
81
site
RJP
name
Ninevite 5
Late third millennium
1
Tell al-Hawa
24
5
0.4
9
Tell Kuran
4.2
6.5
10
2.2
4.5
11
0.7
12
1.5
13
2
0.1
14
3.8
3.8
15
Mowasha
1.3
1.3
18
1.3
1.3
Tell al-Butha 1
6
6
20
2.8
22
23
Tell Wardan
4.2
2.6
0.5
7.2
30
1.9
3.4
37
0.1
0.2
39
Khirbet Gar Sur
1.5
40
19
26
21
25
28
29
42
19
28
43
45
26
66
Khabur
0.1
66
2.8
1.6
4.6
7.2
8.3
0.1
South Garsur
0.1
0.1
5.9
Kharaba Tibn
5.5
17.2
18.3
2
48
2.1
3
6.1
49
1.9
50
1.4
52
0.1
54
0.1
55
57
58
Tell Warada
5
60
1.6
71
Tell Hamide
0.5
72
0.1
73
1.6
74
2
76
79
0.1
0.1
0.4
0.1
1
5.6
0.5
0.1
Table 5.7: Size estimates (ha) for NJS sites with a third–early second millennium bc component (based on Wilkinson and Tucker
1995 and Ball et al. 1989 (Tell al-Hawa), with modifications). Second column shows concordance with RJP site numbers as listed
by the Jazirah Salvage Project (Altaweel 2007: table 1).
82
site
RJP
80
83
name
17
Ninevite 5
Khabur
2.8
0.1
86
Tell al-Hilwa
1
87
2.8
90
Tell Abu Hajira
6.1
6.1
6.1
91
Tell Talab
5
7.5
10
92
Tell Uwaynat
1.5
1.5
3
93
Tell al-Samir
10
19
19
96
Tell Ghubain
1
99
1
108
0.5
Tell Kiber 2
1
121
1
123
al-Kibar
1.8
126
1
127
Abu Kula
131
132
1.1
136
2.1
138
140
Tell Mana’a
1
152
0.5
154
155
Tell Hayal
115
159
23
2
5
10
10
0.9
0.9
0.5
8
0.9
2
2.4
0.7
160
1.6
169
2.2
175
1
177
1
1
Table 5.7: Continued
predict at a relatively early stage whether there will be sufficient rain for crops. Should precipitation
appear insufficient, they can take precautions such as switching failing agricultural land to pasture,
or selling herd animals to buy or trade other foodstuffs (Wilkinson and Tucker 1995).
The North Jazira Survey (NJS) (Fig. 5.1S) was initiated as a side-project of the Tell al-Hawa
excavation, and focused on the immediate hinterland of this large site. The shape of the survey
area was not dictated by archaeological or landscape considerations, but by conservation issues,
because the area was part of a planned large-scale irrigation project. During four seasons, 184 sites
were recorded, several of which were excavated. Site detection was performed by car and on foot
along transects spaced 500 m apart. This allowed the recognition of sites as small as 100 m in diameter and 1 m high. Pedestrian survey to collect sherds was carried out in selected off-site areas.
This strategy was checked against available aerial and satellite images, indicating that only a few
83
250
200
150
100
50
0
Ninevite 5
Khabur
Fig. 5.6: Total number of sites (bars) and aggregate site size in hectares (line) for every NJS period without correction for site
contemporaneity.
155
73
132
154
159
152
160
131
93
175
90
80
91
126
169
72
30
9
28
71
87
10
5
1
37
12
83
76
74
177
29
79
19
86
18
20
22
14 15
13
58
43
42
48
23
26
60
140
40
39
11
54
49
50
45
57
55
52
92
108
96
138
99
115
127
121
123
136
N
0 km 5
10
Fig. 5.7: Map of the NJS region indicating all sites with a Ninevite 5, Late third millennium, or Khabur occupation (after
Wilkinson and Tucker 1995).
sites were missed during the actual survey. Large sites were divided in sampling sectors in order to
detect differences in settlement. These units were either individual mounds, or topographic units
(Wilkinson and Tucker 1995).
Site size estimates are taken from Appendix C: Site catalogue in Wilkinson and Tucker 1995.
Where site size is listed as ‘trace’ in the primary catalogue (Wilkinson and Tucker 1995: 125), it
is here interpreted as an occupation of 0.1 ha. The largest site in the NJS area, Tell al-Hawa, has
been the subject of regular excavations (Ball et al. 1989), and the results are incorporated in the
site’s size estimate. Some of the other sites surveyed by the NJS have been sounded. The results
of these operations, of which only few details are available, seem to have been included in the site
size estimates made in Wilkinson and Tucker 1995. The results of Iraqi excavations within the NJS
area are now being published (Altaweel 2006, 2007), and relevant results will be discussed below.
The resulting dataset is presented in Table 5.7 and Figs. 5.6 and 5.7.
As part of the Jazirah Salvage Project, a number of sites within the NJS area have been sounded by Iraqi archaeologists.49 Excavations at Tell al-Butha 1 (NJS 19/RJP 21) seem to confirm the
results of the NJS for the third and early second millennium bc. The sounding at South Garsur
49 In the first publication on these data, the project was called the Ray Jazirah Project, abbreviated as RJP (Altaweel
2006). In the second publication, the name for the project was corrected to Jazirah Salvage Project, but the abbreviation RJP was retained in the interest of maintaining consistency between publications (Altaweel 2007: 117).
84
(NJS 42/RJP 28) consisted of a single 10x10 m square at the top of the mound. Six levels were
identified, of which the lower four were dated to the Khabur period. In each of these levels, walls
were found, but no conclusions as to the size of the building, or the size of the site as a whole,
could be drawn from these results due to the limited exposure. No changes have been made to the
size estimate based on the results of the NJS. At Tell Kiber 2 (NJS 115/RJP 23), Iraqi archaeologists carried out a surface survey, and excavated four 6x6 m squares. The results of the surface
survey confirmed those of the NJS. No excavated squares yielded only storage pits, suggesting that
the site may have been seasonally occupied during some of the later occupation periods. Whether
this was also the case for the Khabur period is unclear, as the excavated squares were not located
in the area with the highest density of Khabur ware on the surface. Given the fact that the site as
a whole measured 4.5 ha, the original estimate of 1 ha for the Khabur period seems conservative
and will be used here. Two long but shallow trenches were excavated in Tell Hujiera 2 (NJS 130/
RJP 18). Architectural remains were only encountered in one of these trenches, due to significant
disturbance at the site. Some possible Khabur ware sherds were reported (Altaweel 2007: 122), but
they did not come from well-defined stratigraphical layers. The NJS found no Khabur ware at this
site, and it is therefore questionable whether the site was ever occupied during this period. As a
result, it will not be included in the present analysis.
5.3 Reconstructing regional settlement trends
The data presented above can be used to reconstruct the diachronic population development for
each survey region by applying the simulation method that was presented in Section 4.4. Step-bystep application of this method to the BS, B-EDS, and NJS data sets will be discussed here. In
order to estimate the amount of simultaneously occupied hectares per phase, it is necessary to distinguish between different site size classes (see Section 4.4.3). These size classes are derived from
the combined BS, B-EDS, and NJS data sets (Tables 5.3, 5.6, and 5.7). The distribution of site
sizes for this combined data set is shown in Fig. 5.8. Visual inspection of the distribution suggests
70
60
N sites
50
40
30
20
10
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
site size (ha)
10
20
30
40
50
60
70
Fig. 5.8: Distribution of site sizes (ha) across all periods for the combined BS, B-EDS, and NJS data sets (n=368).
site size class (ha)
n sites
mean site size (ha)
standard deviation
0.1–0.3
72
0.2
0.08
0.4–0.9
92
0.6
0.13
1–6
156
2.2
1.3
7–9
22
7.3
0.9
10–30
24
16
6.3
30+
2
individual
0
Table 5.8: Site size classes, number of sites per class, and mean size per class in hectares (n=368).
85
date BC Jezirah
1600
A
B
C
MJ I
1700
1800
1900
OJ II-III
2000
OJ I
2100
EJ V
2200
2300
EJ IV
2400
EJ IIIb
2500
2600
EJ IIIa
2700
EJ II
50
100
150
200
250
20
40
60
LC 5
80
3100
100
EJ 0
20
3000
40
EJ I
60
2900
80
2800
Fig. 5.9: Settlement trends through time. Solid lines indicate the corrected aggregate site size, shaded areas indicate the possible
range of aggregate site size, bars indicate the corrected number of contemporaneously occupied sites, and error ranges indicate
the possible range of contemporaneously occupied sites. (A) B-EDS region. (B) BS region. (C) NJS region.
the following classes: 0.1–0.3 ha, 0.4–0.9 ha, 1–6 ha, 7–9 ha, 10–30 ha, and 30+ ha. The mean site
size for each class is shown in Table 5.8. Within each survey region and within each chronological
phase, each site is assigned to one of these classes, and the simulation method is independently
performed on each of these classes (see Section 4.4.3). By adding up the site counts and site size
estimates per class, one arrives at regional site counts and size estimates for each chronological
phase (Fig. 5.9). This figure also displays the potential minimum number of simultaneously occupied sites, that is the number of type b sites, and the potential maximum number of simultaneously
occupied sites, that is all sites with evidence for occupation within a given phase.
5.3.1 Demographic trends
The total amount of occupied hectares can then be used to reconstruct a diachronic population
development, employing the on-site population density value of 150–250 persons/ha, as determined in Section 4.5 (Fig. 5.10). Fig. 5.10 is presented on an absolute timescale rather than per
phase to allow comparison of these demographic developments with the climate developments
presented in Section 2.3.7.
Between 2900 and 2500 bc, the Birecik-Euphrates area experienced a relatively high demographic growth, followed by a period of stability until 2100 bc (Figs. 5.9A, 5.10A). At that time,
population peaked around 6500–8800 persons. After this stable period, the population of the valley declined rapidly to 6200–7800 persons, with only a few small settlements seeing occupation
during the Late MBA period. The second population explosion during the second quarter of the
second millennium was largely due to the growth of Carchemish. The early third millennium and
the middle second millennium population rises were relatively small when expressed in absolute
numbers. For the region itself, the population growth may be quite marked, but when compared
86
50
100
0.3
150
0
350
-0.3
200
7000
250
14000
-0.2
300
21000
0.3
-0.1
0
42000
-0.6
0
5000
28000
10000
-0.4
35000
15000
0.6
-0.2
0
30000
-0.3
0
2000
-0.2
20000
4000
-0.1
25000
6000
0
D
aggregate site size (ha)
0.2
8000
C
population size
0.4
10000
0.1
12000
MJ I
0.2
1600
B
population size
0.1
A
population size
0.2
date BC Jezirah
0
1700
1800
1900
OJ II-III
2000
OJ I
2100
EJ V
2200
2300
EJ IV
2400
EJ IIIb
2500
2600
EJ IIIa
2700
EJ II
population growth (%)
-0.3
LC 5
-0.2
3100
-0.1
EJ 0
0
3000
0.1
EJ I
0.2
2900
0.3
2800
Fig. 5.10: Reconstructed population levels and population growth through time. Shaded area indicates the population size under
a population density of 150–250 persons/ha. Dashed line indicates population growth rates through time, based on corrected
site counts and site size estimates. Solid line indicates the combined aggregate site size. (A) B-EDS region. (B) BS region. (C)
NJS region. (D) The B-EDS, BS and NJS regions combined.
to the developments in the other regions, it is relatively insignificant. This observation may call
into question the idea that demographic crashes, such as noted for the LRS or the BS, were compensated by synchronous increases elsewhere (Wilkinson 1999: 53). Apart from the fact that the
marked settlement peak that was evident in the reconstruction by Algaze et al. is now much more
fluid, it also does not seem to be exactly in phase with the developments in the Balikh Valley.
Inspection of Fig. 5.9A furthermore suggests that the increase in settled hectares toward 2100 bc
resulted primarily from the growth of already existing settlements. If new people arrived in the
region, it might be expected that they founded new settlements, but this seems not to have been
the case in this area.
For the Balikh Valley, the following demographic trend can be reconstructed (Figs. 5.9B, 5.10B).
From the middle VIab period onwards, and presumably already starting in the Late Chalcolithic,
there was a gradual rise in both the number of sites, and the total amount of hectares that was occupied at any moment in time. This results in a rather steep population growth curve for the first
half of the third millennium bc. The total population of the Balikh Valley peaked around 2400 bc,
when 11000–19000 people lived in the recorded settlements. The Balikh VIcd period witnessed a
steadily declining population, with a minimum of 7700–13000 people at the VIcd–VIIa transition.
Note that, in tandem with the population decline, the number of occupied sites declined as well,
though mean site size increased markedly to almost 4 ha. After this decline there followed a second period of population growth to a level exceeding that of the third millennium. The maximum
population of the Early and Middle Bronze Age was reached at the end of the Balikh VIIb period
87
when 14000–24000 persons lived in the Balikh Valley. Period VIIc was, like period VIcd, a time
of continuous population decline. This trend continued well into Balikh period VIII and was only
halted and reversed during the Middle Assyrian period (Lyon 2000).
In the NJS region, population decreased slightly during the early Ninevite 5 period to a minimum of 9300–15000 people (Figs. 5.9C, 5.10C). This absolute minimum provided the basis for a
continuous growth lasting some one thousand years and covering much of the third millennium
and the early part of the second millennium. During the Khabur period, an estimated peak population of 25000–42000 persons lived in the settlements of the NJS region. During this period, the
large central site of Tell al-Hawa accounted for almost 40 % of the total population living in the
area. During the second quarter of the second millennium bc population dwindled, but remained
above the low level recorded for the Ninevite 5 period. In this respect, the NJS population curve
differs from the BS and the B-EDS curves, which show decreases at the third–second millennium
transition. For this period, the NJS growth curve levels off slightly, but there is no evidence for
any population decline during this period.
Several researchers have used reconstructed population growth rates to detect periods of
anomalous growth and/or decline and to detect periods of population displacement (Wilkinson
2000a; Ristvet 2005: 198–9). The percentual rate of population growth r% is calculated as (Hassan
1981: 139):
(3) r% = ( ln ( N2 / N1 ) / t) ∙ 100
where N2 is the population reached after t years from an initial population N1. The results of this
calculation are presented in Fig. 5.10A–C. It is immediately apparent that population growth fluctuates widely, ranging from very high growth rates to rapid declines. Average growth rates for past
societies are estimated at 0.04–0.16 % by Hassan (1981) and at no higher than 0.1 % and 0.13 %
by Maisels (1990: 126) and Carneiro and Hilse (1966: 179), respectively. In the surveys analysed
here, periods with an average growth rate are rare. These periods include the late VIab, late VIIa,
and late VIIb periods of the BS region, the early part of the Middle EBA period of the B-EDS
region, and the entire Late third millennium and early part of the Early second millennium periods
of the NJS region.
However, closer inspection of the population growth rates suggests that there seems to be a
positive correlation between demographic stability and population size. In other words, the region
with the largest population, that is, the NJS region, seems to have experienced the highest stability
in terms of population growth and decline, and the population growth rate remains closest to the
long-term average growth rate. When the population developments in each region are combined,
the population growth rate becomes even more stable and generally falls within the assumed range
of natural growth (Fig. 5.10D). There was one minor decrease around 2000 bc, and a significant
decline only set in around 1700 bc.50
At first sight, the observation that the combined demographic graph shows a more stable
population development than the regional graphs could indeed lend support to the theory that local population deficits in one region were offset by incoming populations from elsewhere, which
would result in the maintenance of a long-term demographic balance (Wilkinson 2000a: 242).
However, it should be realized that net population change, as reflected in Fig. 5.10, is the difference between the number of births per year and the number of mortalities per year, which were
probably quite high. Relatively minor changes in either the fertility or mortality rate can therefore
potentially completely transform the net population growth rate. Such changes can result from
changes in inter-birth spacing, child mortality, average age of first pregnancy, and so forth (Hassan
1981: 126). If the birth rate dropped below the mortality rate, the population would decrease without any external interference (Cowgill 1975: 512). This suggests that local demographic trends may
have been largely driven by local developments and changes in fertility and mortality rates and that
there were no cross-regional processes at play.
50 Fig. 5.10D only displays the combined trends for the period for which data from all three surveys is available, i.e.
2650–1550 bc.
88
5.3.2 Rank-size graphs
Rank-size plots for the three survey regions were constructed according to the method outlined in
Section 4.6 (Figs. 5.11–13). Because the exact number of settlements, and more importantly their
sizes, is only known for the transition points between periods (see Section 4.4.3), rank-size plots
have only been constructed for these specific points in time.
The rank-size plots of the B-EDS settlement system are constructed on the basis of very few
sites, making any observations made from them highly suspicious (Fig. 5.11). The primate distribution that is observed for all periods, however, probably indicates that the local settlements always
belonged to the hinterland of a larger site. The 2200 bc rank-size distribution is slightly reminiscent
of a primo-convex distribution (Fig. 5.11B). At least from the mid-third millennium, this primate
site was most probably Carchemish.
During the early third millennium, the settlement pattern of the BS region conforms almost
perfectly to the log-normal distribution (Fig. 5.12A). This log-normality decreases in favour of a
more convex distribution, reaching highest convexity at c. 1800 bc (Fig. 5.12B, C). After that point,
convexity seems to decrease again (Fig. 5.12D, E), but there remain large error margins, making
any more specific observations impossible. In fact, apart from the observation that the 1800 bc
and 1700 bc settlement patterns were very probably convex, no other observations can be made
with any measure of confidence.
The NJS rank-size plots are again constructed from a reasonably large data set, making conclusions drawn from them more certain (Fig. 5.13). At the beginning of the third millennium, there
was a slightly primate or possible primo-convex settlement pattern (Fig. 5.13A). Although the large
error margins make any observations problematic, there seems to be a very slight development
towards a more log-normal distribution during the late third and early second millennium (Fig.
5.13B–D). The overlapping error ranges for the A values of different periods again make any more
detailed interpretation hazardous. As expected, the rank-size plots of the NJS region compare
reasonably well with those produced by Wilkinson and Tucker (1995: fig. 52) for the uncorrected
settlement record of the third and early second millennium bc.
100
A
100
B
size (ha)
10
size (ha)
10
1
0.1
1
1
10
rank
100
1
10
rank
100
0.1
1
10
rank
100
100
C
size (ha)
10
1
0.1
Fig. 5.11: Rank-size plots for the B-EDS region at (A) 2500 bc, (B) 2200 bc, and (C) 1900 bc, based on corrected site counts and
site size estimates. Dashed line is log-normal distribution. Grey zone represents 90 % confidence range.
89
100
A
100
B
size (ha)
10
size (ha)
10
1
0.1
1
1
10
rank
100
C
0.1
100
1
10
rank
100
1
10
rank
100
100
D
size (ha)
10
size (ha)
10
1
0.1
1
1
10
rank
100
1
10
rank
100
0.1
100
E
size (ha)
10
1
0.1
Fig. 5.12: Rank-size plots for the BS region at (A) 2400 bc, (B) 2000 bc, (C) 1800 bc, (D) 1700 bc, and (E) 1550 bc, based on corrected site counts and site size estimates. Dashed line is log-normal distribution. Grey zone represents 90 % confidence range.
As argued in Section 4.6.1, reliable rank-size plots can only be constructed if the limits of the
settlement system on which the analysis is performed are correctly identified. Failure to do so may
result in a convex distribution because more than one system is analysed, whereas a primate distribution is likely to emerge when only part of a larger settlement system is included in the analysis
(Johnson 1977: 498). Comparison of the B-EDS, BS, and NJS suggests that the BS region is least
likely to have suffered from these effects, and the NJS most likely. Although the BS focused on the
river valley, the survey’s western and eastern boundaries were the steppe area where no more sites
were encountered (Akkermans 1993: 139). This boundary may have equally applied to the south,
where Tell Bi’a and its immediate surroundings, as well as the Euphrates, may have represented
a natural boundary prohibiting settlement further south. Only the Syrian–Turkish boundary represents a truly artificial boundary to the survey area (but see Akkermans 1993: 139). The B-EDS
was bordered to the south by the Turkish–Syrian border, to the north by Halfeti, and up into the
highlands by an artificial line drawn at 400 m above sea level. In fact, the B-EDS covers the northern half of the so-called Carchemish sector (Peltenburg 2007a: 6–7). Finally, the borders of the
NJS were dictated by the fact that the area was to become part of a large-scale irrigation project,
and therefore apparently did not fully encompass any natural archaeological or landscape zone.
The rather arbitrary nature of the borders of the B-EDS and NJS regions make the results of the
rank-size analyses for these regions suspect, and any possible effects will be evaluated in Sections
8.2.1 and 8.2.3.
90
100
A
100
B
10
size (ha)
size (ha)
10
1
0.1
1
1
10
rank
100
C
0.1
100
1
10
rank
100
1
10
rank
100
100
D
size (ha)
10
size (ha)
10
1
0.1
1
1
10
rank
100
0.1
Fig. 5.13: Rank-size plots for the NJS region at (A) 3000 bc, (B) 2500 bc, (C) 2000 bc, and (D) 1500 bc, based on corrected site
counts and site size estimates. Dashed line is log-normal distribution. Grey zone represents 90 % confidence range.
5.4 Contextualizing the reconstructions
Section 5.4.1 discusses the evidence from other surveys in order to place the present reconstructions in perspective. The nature of the primary data does not allow the reconstruction of continuous population trends but permits the detection of very general trends, besides touching on issues
of scaling and settlement patterning across the landscape. Section 5.4.2 discusses contemporary
historical texts that provide information on settlement sizes and organization.
5.4.1 The results from other surveys
Apart from the surveys that have been analyzed here, numerous other surveys have been conducted in the Jezirah (Fig. 5.1). However, many of these surveys have not been published the
level of detail necessary for the reconstruction that is attempted here. As a result, the conclusions drawn from these surveys are often more general. Nevertheless, their results can be used to
point out large-scale trends and contextualize the current reconstruction. In other words, these
surveys can help determine the degree to which the trends identified here are typical for northern
Mesopotamia as a whole. The uncorrected settlement trends documented by these surveys are presented at the end of this section in Fig. 5.14.
The Syrian Upper Euphrates settlement system, comprising the extensively researched areas
of the Tabqa and Tishrin Dams, has been evaluated by several researchers (Peltenburg 2000;
Wilkinson 2004; Cooper 2006; Peltenburg 2007c; Schwartz 2007) (Fig. 5.1B, C). In a sense, it is
a continuation of the B-EDS region described above, which has only relatively recently become
separated by political borders. This political situation perhaps influenced archaeological research
in both areas. Thus, a recent volume on the Upper Euphrates stopped at Carchemish (Cooper
2006), and the present study also uses the large site on the border between Turkey and Syria as an
artificial border between two zones that are reviewed in a different context. On the other hand,
the intensity of research in both areas is comparable, as the research in both cases derived from
dam construction resulting in the permanent loss of cultural heritage. Furthermore, Peltenburg
has suggested that the area around Carchemish indeed represented a transitional environmental
91
zone, which may have influenced cultural traditions as well (2007a: 8–9). This gap between two
different constructed culture areas is only recently being bridged with the publication of a volume on Carchemish, which focuses on its unique cultural traits and the contacts with both the
north and the south (Peltenburg 2007c). The results of the surveys in the Tabqa Dam area were
compiled by Reichel (2004), who also included data from the numerous excavations in the region.
Wilkinson (2004: table 7.1) estimated an aggregate occupation size of 80–90 ha for the combined
Hadidi–Sweyhat settlement system during the middle and late EBA, which was a considerably
larger area than that of the early EBA, due mainly to the growth of the two regional centres Tell
Hadidi and Tell es-Sweyhat. During the early second millennium, large parts of the Sweyhat plain
were abandoned, including Tell es-Sweyhat itself. Schwartz (2007) also documented the apparent
wide-spread abandonment of the Tabqa and Tishrim Dam areas, with many sites being at least
temporarily abandoned. At other sites, there was evidence for destruction, directly followed by reoccupation (Schwartz 2007: figs. 4, 5).
The first season of the Land of Carchemish survey was carried out in 2006 (Fig. 5.1D).51 The
survey targeted the triangle formed by the Sajur River, the Euphrates, and the Syrian–Turkish border. In 2006, the western part of this area was investigated. Although parts of this triangle had
been surveyed before, there still was an incomplete and unbalanced settlement record focusing on
tell sites along the Euphrates, and this survey aimed to correct this bias. During the first season,
twenty-three sites, typologically divided in small, flat sites, tell-sites, and hilltop sites, were recorded. During the third and early second millennium bc, tells dominated the settlement record. Two
sites contained Early EBA material, six sites yielded Middle/Late EBA material, and MBA material
was also found at six sites. Four out of six sites with M/L EBA material also had MBA material,
suggesting a prolonged settlement preference for these locations (Wilkinson et al. 2007b).
In 1991–2, the Westgazira survey (Fig. 5.1E) targeted the northern half of the upland area
between the Euphrates and the Balikh, up to the Syrian–Turkish border.52 Most sites were located
in the Saruj plain in the northwest corner of the survey area, and along the Wadi Qaramogh. EB
ceramics were found at numerous sites, including the large sites of Tall Kufaifa, Tall Matin, and
Boz Höyük. At Tall Kufaifa and Tall Matin, stone city walls were observed. At Tall Matin, the stone
layout of a temple in antis was recorded, suggesting a mid–late third millennium bc date for this
site, and possibly for the city wall as well. The area also seems to have been relatively densely settled during the MB period, but only a few details on this period are given (Einwag 1993).
The Balikh-Euphrates Uplands survey was carried out in 1996 and 1997 and recorded twentyeight sites in this area, dating from the early third millennium bc to the early twentieth century ad
(Fig. 5.1F).53 The majority of these sites dated to the Early Bronze Age or to the Roman to early
Islamic period. Four sites were dated to the third millennium, these being Tell al-Hassan/Tell Jedi
(site 14), Khirbet Taha (site 17), Joub esh-Shayeer (site 22), and Tell Shayeer (site 23). Khirbet Taha
was a 4 m high, c. 3 ha mound with clear evidence for early third millennium bc occupation. Tell alHassan/Tell Jedi was located along a wadi and estimated at 5–10 ha. Stone foundations were dated
to the mid to late third millennium bc. Apparently, this was the only period of occupation of this
site, although a transitional EB/MB phase may also have been present. Joub esh-Shayeer was also
dated to the mid–late third millennium bc, and was again located along a wadi. The site consisted
of several small 0.25 ha mounds. Tell Shayeer was dated to the same period. This site consisted of
a 5 m high, 2–3 ha conical mound surrounded by a lower tell. In both areas, mid–late third millennium bc material was abundantly present. In so far as any conclusions can be drawn from these
four sites, it seems that occupation in this region was more intensive during the mid–late third mil51 The survey employed a combination of survey by car, transect walking, and recording of sedimentary profiles.
Emphasis was placed on locations that were difficult to access by car, i.e. hill tops and wadis.
52 The survey was carried out in three short seasons by a very small group operating from Raqqa. Sites were mapped, and
diagnostic ceramics were collected. Only tell-sites seem to have been visited. Several sites along the Wadi Qaramogh
were also visited (Einwag 1993).
53 Sites were identified on maps and aerial/satellite imagery. The area between the Euphrates and the Balikh, bordered
on the north by the region already investigated by Einwag (1993), was then divided in three different environmental
zones, and within these zones, a sample of 5x5 km squares was investigated. Actual fieldwork involved both intensive
field walking and driving around by car. In the end, it turned out to be impossible to achieve these objectives, and only
squares within two out of three environmental zones were investigated (Danti 2006–7).
92
lennium than during the early part of the Early Bronze Age, both in terms of the number of sites
and their sizes. Sites were preferably located along wadis, indicating a concern for water supplies
(Danti 2006–7).
The Euphrates Valley around Tell Bi’a was surveyed by Kohlmeyer in 1983 (Fig. 5.1G).54 Sites
were recorded on both sides of the Euphrates. There were very few large, continuously occupied
tell-sites in this area of the valley. This absence may have been caused by the need to occasionally
relocate settlements to new locations, as dictated by shifts in the course of the Euphrates. The
team found many one-period sites, suggesting that many more have been covered by Euphrates
sediments. The EB period may have been among the most densely occupied periods in this area.
The large EB site of Tell Aswad was located 15 km southeast of Tell Bi’a. This site had a diameter
of c. 550 m, and thick ash deposits were found in the sections of a canal cut through the site. A site
of similar size was found 12 km west of Tell Bi’a. There are several sites with evidence of stone
city walls, especially toward the narrow gap of Halabiye–Zalabiye. A few EB sites show occupation
continuity into the second millennium (Kohlmeyer 1984).
A survey of the Harran plain has indicated that the area was intensively occupied during many
prehistoric as well as historic periods (Fig. 5.1I). Important sites in this area include Harran itself,
occupied from at least the late third millennium bc, as well as Kazane Höyük. Given the proximity to the Balikh Valley, an occupation pattern similar to that recorded there might be expected,
but this suggestion remains purely speculative. Unfortunately, apart from a preliminary report
(Yardimci 1993), no details of this survey have as yet been published.
The area of the Jebel Bishri up to the Euphrates was recently surveyed by a Finnish team in
order to study ‘how the pastoral nomadic life has developed in the region in relation to agricultural
village life from ancient times to the present’ (Lönnqvist 2006: 207) (Fig. 5.1J). 55 The area is of
particular interest here because it has long been seen as the homeland of the Amorites (see Section
7.2). Today, the area receives less than 150 mm annually, and farming is only occasionally possible
in some wadis. Even during climatically favourable periods, the area remained relatively marginal
and for example never developed a continuous tree cover. The area yielded abundant evidence for
occupation starting as early as the Middle Palaeolithic period. Some stone enclosures were associated with Early Bronze Age pottery. Cairns/tumuli located on hilltops were also associated with
Chalcolithic/Early Bronze Age pottery, although it should be noted that these features are still in
use today. At one site, an Ubaid sherd, pieces of a holemouth jar, and Early Bronze Age and possibly Middle Bronze Age pottery was found. The terraces bordering the Euphrates Valley yielded evidence of more permanent settlement, including tell-sites. This region also showed more evidence
of occupation during the Middle Bronze Age, as attested by sherds with comb-incised straight and
wavy lines. It seems that the desert-steppe area was most intensively used during the Early Bronze
Age and in the Roman period (Lönnqvist 2006). This conclusion reinforces the picture that had
already emerged from the survey of the Euphrates Valley directly north of the Jebel Bishri.
The Wadi Hammar Survey was carried out between 1997 and 2000 within a 10 km radius around
the site of Kharab Sayyar, near Tell Chuera, and thus provides contextual data for both the BS
and the Upper Khabur Survey (Fig. 5.1K). Sixty-five sites dating to all periods were located by the
survey team.56 There were apparently no Late Chalcolithic/Uruk sites, but a period of apparently
low activity was followed by a settlement peak during the third millennium, when eleven sites were
occupied. Among these sites was Kharab Sayyar, which now seems to have been occupied during
54 Sites were marked on a 1:25000 map, and also mapped at a 1:5000 scale. No aerial imagery was available. Larger sites
were divided in sampling units according to site morphology.
55 The survey made extensive use of aerial/satellite imagery to construct a Geographic Information System (GIS) of
the region into which all sites and features were plotted using GPS. Small areas were selected for field walking in lines
spaced 15 m apart. There was a specific focus on sites, being loci of past human activity, rather than on documenting
the continuous sherd and flint blanket covering the landscape, as it was argued that this blanket resulted from erosional
activity. The survey also utilized ground penetrating radar in selected areas (Lönnqvist 2006).
56 Details on survey strategy are unavailable, but the survey seems to have been site-based. Of the sixty-five recorded
sites, 50 % was found by the survey team, 40 % was found using information provided by the local population,
and 10 % was detected on satellite images during the preparation for the survey (http://web.uni-frankfurt.de/fb09/
vorderasarch/survey.htm (Accessed 16 November, 2007)).
93
the entire third millennium (Hempelmann 2008). Only two sites were occupied during the Middle
Bronze Age (2000–1600 bc). For the Late Bronze Age (1600–1200 bc), ten sites were recorded
(http://web.uni-frankfurt.de/fb09/vorderasarch/survey.htm (Accessed 16 November, 2007)).
The Upper Khabur Survey (UKS) aimed at a longue durée population reconstruction of the
Upper Khabur Basin (Fig. 5.1L). It was originally envisaged to cover the entire Upper Khabur
Basin, but only the western part was surveyed as part of this project. The survey covered a roughly
triangular shape that was bordered to the west and southwest by the Khabur, to the north by the
Turkish–Syrian border, and to the east by the Wadi Jaghjagh.57 Much of this material is at the stage
of analysis and publication; so far the early periods have been published, as well as the second and
early first millennium bc material (Lyonnet 1996, 1998, 2000, 2004; Faivre n.d.). A closer inspection of the adopted survey strategy suggests that, although some measures have been taken to create a representative sample of sites, at the very least small sites that are neither indicated on maps,
nor clearly visible in the field, may be severely underrepresented. The focus on potential second
millennium sites further suggests that the survey was likely to enhance the picture of second millennium settlement in the region, rather than result in new insights in local settlement patterns.
With these caveats in mind, the following settlement development can be sketched (Lyonnet 1996).
From the late fourth to early third millennium, the settlement pattern in the western Khabur seems
to have been relatively stable, although both the number of sites and the total amount of settled
hectares increased considerably. During the Ninevite 5 period, there seems to have been a cultural
divide between the northeastern and southwestern part of the survey region. Incised and painted
wares were only recognized in the northeastern part, but it is unclear to what degree this cultural
divide was also present in non-decorated wares (Lyonnet 1996: 368). During the second part of
the third millennium, and more specifically the later part, the number of settlements and their
size decreased again, reaching its lowest values during the early second millennium. At that time,
the course of the Khabur north of Hassakeh was completely devoid of permanent settlements.
Lyonnet interpreted the sites in that region as pastoral nomadic campsites, based on the absence of
Khabur ware in this region, and the presence of a certain type of cooking pot (1996: 371). During
the Mitannian period, the number of settlements and their aggregate size grew again. Note that
this picture is considerably distorted by the fact that several important sites in the wider region of
the western Upper Khabur were not surveyed by Lyonnet, including Chagar Bazar, Tell Ailun, and
notably Tell Mozan. Tell Mozan alone would have had a larger population than all the other sites in
the Upper Khabur Survey region combined, if the size estimate for Tell Mozan of 120 ha during
the late third millennium bc is correct (Pfälzner and Wissing 2004: 81). Furthermore, considerable
discrepancies exist between Lyonnet’s size estimates and those made by the Tell Beydar Survey
(Ur 2004). The evidence from the Tell Beydar Survey and from excavated sites, for example Tell
Beydar (UKS 15/TBS 1) and Tell Arbid (UKS 42), suggests that Lyonnet’s estimates were probably relatively low.58
57 The Upper Khabur Survey was carried out in 1989 and 1990. Before the actual fieldwork, sites were located on
1:200000 maps from the French Mandate period. No aerial or satellite photographs were used. Over 300 tells were
located, of which 63 were actually surveyed by Lyonnet, and several more by Nishiaki who primarily focused on
Neolithic and earlier occupations. The choice for specific sites to be surveyed was made in advance and based on (a)
site location so as to ensure a representative regional sample that included sites located in a variety of environmental
contexts, (b) site size, i.e. all large sites were included, (c) previous knowledge of sites, i.e. a focus on less well-known
sites, and (d) a focus on sites where second millennium material could be expected. In the field, some small sites not
indicated on the map were included as well. Sites were divided in sampling units according to site morphology. These
units were separately sampled in order to allow recognition of shifting on-site settlement and activity areas (Lyonnet
2000).
58 Sites that were surveyed by the Upper Khabur Survey as well as the Tell Beydar Survey included Tell Beydar (UKS 15/
TBS 1), Tell Khatoun (UKS 16/TBS 32), Tell Effendi (UKS 17/TBS 55), Tell Hassek (UKS 28/TBS 43), Tell Berguil
Bouz (UKS 30/TBS 22), Tell Rhazal Tahtani (UKS 51/TBS 63), and Tell Boughaz (UKS 52/TBS 65, there called Tell
Khazna). The results from the Tell Beydar Survey differed in some important aspects from the Upper Khabur Survey.
Notably, the Upper Khabur Survey revealed substantial early third millennium bc occupation at Tell Jamil (UKS
14/TBS 59), Tell Khatoun (UKS 16/TBS 32), Tell Berguil Bouz (UKS 30/TBS 22), and Tell Rhazal Tahtani (UKS
51/TBS 63), identified by a non-decorated pottery assemblage. The Tell Beydar Survey found no evidence at these
sites for their period 6, i.e. the early third millennium bc. Note, however, that early third millennium diagnostics are in
any case rare in the Tell Beydar Survey, possibly the result of the intensive collection strategy of the earlier Upper
94
The Tell Beydar Survey (TBS) was carried out in 1997 and 1998 and covered a 452 km2 circular area centred on Tell Beydar (Fig. 5.1M). This area included a part of the large basalt plateau to
the southwest of Tell Beydar where only very few sites were recorded. 59 Overall, eighty-three sites
with a total size of 246 ha were found. Sites were preferably located along or nearby watercourses.
Periods 06–08 covered the third and early second millennia bc. During the early third millennium
(period 06), there were twelve sites within the Tell Beydar Survey area covering 45.3 ha, of which
17.0 ha was taken up by Tell Beydar itself. Like the Upper Khabur Survey, the Tell Beydar Survey
only infrequently found Ninevite 5 sherds. During the mid–late third millennium (period 07),
twenty sites were occupied, including all period 06 sites, with a total size of 69.6 ha. Beydar was the
largest site at 17.0 ha, but there were also three medium-sized tells covering 7–8 ha each. At sites
where period 07 ceramic material was found, it represented the majority of the collected sherds,
except for a few sites that were subsequently interpreted as representing seasonal occupations. The
close association of period 07 sites with radial lines was also noted. During the early second millennium (period 08), only one site seems to have been permanently occupied, whereas five others
may have been used on a seasonal basis (Ur 2004).
The course of the Wadi Jaghjagh south from Tell Barri, and the area around the confluence
of the Wadi Jaghjagh and the Wadi Radd yielded abundant evidence for occupation (Eidem and
Warburton 1996) (Fig. 5.1O). In this area and around Tell Brak, fifty-six tells were visited of which
twenty-nine yielded third or early second millennium material. Of these, fifteen yielded Ninevite
5 material, twenty-five late ED/Akkadian sherds, and nineteen early to mid-second millennium
material. It is difficult to draw any conclusions from this material (Eidem and Warburton 1996:
55). Site counts alone are not good indicators of changing human activity, and there are reasons
to assume the presence of a significant bias due to the survey strategy adopted. 60 It should further
be noted that most sites are located along the Wadi Jaghjagh, which was a perennial stream before
the advent of mechanized irrigation works. Fielden’s survey of the Jaghjagh between its confluence with the Wadi Radd and Hassakeh yielded the observation that after an Uruk settlement peak,
there was a decline in the number of settlements during the third and second millennia bc, and
settlement was limited to the larger sites (1978/9: 172).61
Several survey seasons were carried out around Tell Leilan between 1984 and 1997 to place the
excavation at Tell Leilan in a regional context (Fig. 5.1P). In 1984, a low-intensity site-based survey
was carried out within a radius of 15 km from Tell Leilan (Weiss 1986). A more intensive survey
in 1987 included an intensive field walking component along the Wadi Jarrah, and controlled site
surface survey of selected sites with evidence for Leilan period II and III occupations (Stein and
Wattenmaker 1990). In 1995 and 1997, as part of the Leilan Regional Survey (LRS) two further
field seasons were carried out that enlarged the survey area to a 30 km wide transect centred on
Tell Leilan, and running from the Syrian–Turkish border in the north across the Wadi Radd to
the Syrian–Iraqi border in the south. South of the Wadi Radd, survey coverage was limited.62 The
following observations on the local settlement history are mainly drawn from this fieldwork. The
Leilan Regional Survey collected material from the Pre-Pottery Neolithic to the late Ottoman period. For the third and second millennia bc, some of this material is now presented in a preliminary
fashion by Ristvet (2005, 2008). Ristvet provided a catalogue of third and second millennium sites
59
60
61
62
Khabur Survey. Site size estimates also vary between the Upper Khabur Survey and the Tell Beydar Survey. The UKS
estimates Tell Effendi (UKS 17/TBS 55) at 1.4 ha and Tell Hassek (UKS 28/TBS 43) at 1.32 ha, whereas these sites
are estimated by the Tell Beydar Survey at 8.5 and 7.4 ha, respectively (Ur 2004).
Toponyms, maps, and Corona satellite imagery were used to identify sites before actual fieldwork. On-site collection
took place in discrete collection units to allow the recognition of shifts in on-site settlement through time.
The fact that all sites are located along wadi courses results from the adopted survey strategy, which was to walk along
the wadis and visit all visible tells (Eidem and Warburton 1996: 58). In so doing, many small sites will certainly have
been missed.
This information will certainly be updated once data from a recent survey in a 15 km radius around Tell Brak under
the direction of H. T. Wright (Fig. 5.1N) become available (Oates 2004; Lawler 2006: 1460–1).
During this survey, satellite imagery and a wide range of maps were employed. Sites were mapped using GPS and a
Total Station. Sites were divided into collection units based on the morphological characteristics of each site. During
the 1995 season, so-called hollow ways detected on satellite imagery were investigated, and a geomorphological study
of several bulldozer cuts was carried out as well (Ristvet 2005).
95
with information on the size of the tell and the number of sherds by which chronological phases
are represented at each site (2005: appendix 3). In the process, three different classes were distinguished. The class 1.000 indicates the presence of more than four sherds definitely assignable to
that phase, 0.001 indicates fewer than four sherds certainly dating to that phase, and 0.0001 indicates ‘the presence of sherds that may date to each phase’ (Ristvet 2005: 310). Assigning sites to
the first two classes seems rather straightforward, but the latter class is rather problematic in that it
does not allow distinction between phases to be made. This suggests that if a phase at a site is classified as 0.0001, there are actually no sherds that could be dated to that phase with certainty, unless
there are other arguments to do so. The amount of occupied hectares per phase is calculated on
the basis of these sherd classes, but how these sherd densities were converted into individual size
estimates, which in fact are not given, is unclear.63
Ristvet distinguished eight chronological phases in the Leilan Regional Survey material for the
period 3000–1500 bc. During phase 1 (Leilan IIIa), seven sites were occupied at a total of 31 ha,
almost half of this area being taken up by Tell Leilan. During phase 2 (Leilan IIIb–c), there was a
considerable increase in both the number of sites and their aggregate size, although the southern
limit of settlement moved slightly north. Ristvet inferred a three-level settlement hierarchy centred
on Leilan (LRS 1), Dogir (LRS 16), and Mohammed Diyab (LRS 55). The number of settlements
declined to twenty-six during phase 3 (2650–2500 bc). At the same time, Leilan reached a size of
90 ha, and there was strong nucleation into large settlements. The number of sites and their aggregate size increased into phases 4 (2500–2300 bc) and 5 (2300–2200). However, the organization of
the settlement system changed slightly, with primary centres except Tell Leilan decreasing in size,
interpreted by Ristvet as centralization on Tell Leilan. During phase 5, the area around the Wadi
Radd was also settled. Phase 6 (2200–1900 bc) coincided with Leilan IIc and was characterized by
a strong reduction of the number of sites and their aggregate size by 69 % and 72 %, respectively.
Average settlement size declined as well, and settlement withdrew north of the modelled ancient
400 mm isohyet. During phase 7 (Leilan I), there were 157 sites totalling 767 ha. Over 75 % of
these sites were less than 5 ha, in size a significant change from the previous urban-based settlement system. However, Ristvet also observed that correction for the site contemporaneity problem indicated that only twenty-eight sites were occupied at the same time, and this observation
severely limits the significance of the Leilan I resettlement. Finally, during phase 8 (1700–1500 bc),
the region seems to have been depopulated again, with site numbers and total site size only slightly
higher than they were during the phase 6 collapse (Ristvet 2005).
The Tell Hamoukar region was surveyed in 2001 and 2002 (Ur 2004) (Fig. 5.1Q). The survey
identified sixty-one sites in a 125 km2 area around Tell Hamoukar.64 Period 06, that is the early third
millennium, is divided in a ‘pre-urban’ and a Late Ninevite 5 phase. Four sites are recorded for each
phase, but total site area increases from c. 10 ha for the pre-urban phase to c. 100 ha for the Late
Ninevite 5 phase. However, late excised Ninevite 5 ware is only found at Hamoukar itself, suggesting that the other sites in the THS region had been abandoned by that time (Ur 2004: 158). During
period 07, the mid–late third millennium bc, all previously settled sites remained occupied, and
two more became settled. Total settled area increased slightly. Apart from these six sites, material
diagnostic of period 07 was found at numerous other sites. It may be assumed that the majority of
this material resulted from off-site manuring practices (Ur 2004: 159). In a number of cases, sherds
were less abraded than might be expected had they resulted from manuring, and these cases could
63 The low numbers of sherds involved in classes 1.000 and 0.001, i.e. more or fewer than four sherds, respectively,
suggests that only very specific types were considered to be representative for each phase, given the fact that normally
hundreds of sherds can be collected from a site. Alternatively, it is possible that the pottery chronology itself is not
detailed enough to allow the assigning of large quantities of feature sherds to specific phases. This possibility is hinted
at by Ur, who observes that it is presently impossible to accurately assess the validity of the Leilan chronology, because
it remains poorly understood (2004: 71). It should also be noted that there are some inconsistencies as regards the
number of sites per period between the data published by Ristvet, and those published on the Tell Leilan website
(Weiss 2007).
64 Sites were identified from Corona satellite images and maps and then visited in the field by vehicle. It was decided not
to identify sites in the field through systematic transect walking, and where fieldwalking was undertaken to document
off-site sherd scatters, no new sites were found. Sites were plotted using GPS and each site was sampled in distinct
units allowing the reconstruction of diachronic on-site settlement changes (Ur 2004: 127)
96
represent pastoral encampments or temporary settlements of seasonal labourers who worked at
Hamoukar during periods of high agricultural activity, for example during harvests (Ur 2004: 160).
Period 08 settlement, the early second millennium bc, was represented at nine sites totalling 25.9
ha. These figures indicate a significant decline of the total population in the Hamoukar area, and
also a change in the organization of the settlement pattern. During the late third millennium, there
were few large sites, while during the early second millennium there were more sites with a smaller
average size.
Meijer (1986, 1990) conducted a survey in the area east of the Wadi Jaghjagh (Fig. 5.1R). 65 To
the north and east it was delimited by the borders with Turkey and Iraq, respectively. To the south
it was delimited by the road from Tell Brak to ‘Ain Diwar. The purpose of this survey was to provide a pottery sequence for the region, and to provide a more solid basis for the interpretation
of Old Assyrian textual sources. The survey partly overlapped with other surveys discussed here,
that is the LRS and the 1988 survey around Tell Brak. Nevertheless, the figures resulting from this
survey are difficult to interpret, due to uncertainties in the ceramic chronology, and the absence of
reliable site size estimates.66 Meijer recorded 16 EB I sites, 12 EB II sites, 55 EB III sites, 50 EB IV
sites, 176 MB sites, and 38 LB sites. The assignment of sites to the MB period occurred primarily
on the basis of the presence of Khabur ware, which has since been shown to occur already during the final centuries of the third millennium bc. Thus, these MB settlements were spread over a
much larger chronological period than any of the other periods distinguished in this survey. Closer
inspection of these figures furthermore suggests that the degree of settlement continuity between
the MB period and the periods bracketing it is relatively low; indicating that the actual number of
contemporaneous sites was also low. Especially the MB sites are primarily one-period sites, indicating a relatively high degree of settlement discontinuity, and an unstable settlement system as a
whole. Thus, the MB settlement peak was probably considerably lower, and may not have been as
pronounced as the raw data suggest. This observation is in line with the reconstructed and corrected settlement trends of the partly overlapping LRS.
The lower and middle Khabur region shows clear evidence for the waxing and waning of human settlement through the ages. An extensive survey and numerous excavations focusing on third
millennium sites provide information on third and second millennium bc human activity in the
region (Fig. 5.1T).67 The following picture emerges from the survey. During the third millennium,
settlements were primarily located along the upper stretch of the Khabur, above 36° latitude.
There were some larger, potentially urban sites just south and east of the Jebel Abd al-Aziz, that
is, outside the valley of the Khabur. The surveyors suggested that some small irrigation canals may
have been constructed in this region, but that it is unlikely that the large regional canal indubitably
attested for the Late Bronze Age was already in existence. 68 It is, however, observed that EB I assemblages in particular may have been underrepresented in the survey (Röllig and Kühne 1983:
194). During the Middle Bronze Age, the settlements outside the river valley disappeared, and the
number of sites along the Khabur clearly diminished in number as well as size. There seems to be
no evidence for local irrigation systems during this period (Ergenzinger et al. 1988: 117–20).
65 The survey started by locating major sites, followed by visiting smaller sites in the vicinity. No aerial photographs were
used. The local population was asked for information. Pottery and other finds were collected in an unsystematic way.
66 The chronological sequence suffered from a lack of well-published and reliable local pottery sequences at that time
(Meijer 1986: 32; cf. Ur 2004: 55).
67 The survey of the Tübinger Atlas des Vorderen Orients was carried out in 1975 and 1977 and focused on the valley of
the Khabur from the Khabur-Euphrates confluence in the south to Hassakeh in the north. The survey focused on
tell sites, most of which were already known prior to the survey. Sites were divided in collection units based on their
shape, and sherds and artefacts were collected from the surface. In total, 129 sites were mapped and/or visited (Röllig
and Kühne 1977/8, 1983).
68 Note, however, that Göyünç and Hütteroth doubt the past existence of a single canal running along the Middle and
Lower Khabur, and argue that it is more likely that there were multiple smaller canals that each served a local irrigation
system (1997: 75–6).
97
M
N
O
P
Q
R
S
T
U
V
43 sherds on 17 sites
no evidence for settlements
24 sites
46 sites (229 ha)
few small (seasonal?)
villages
few
sites?
no evidence for settlements
many large
villages,
Kranzhügel
24 sites (146 ha)
176 sites
50 sites
55 sites
9 sites (26 ha)
15 sites (120 ha)
4 sites (100 ha)
19 sites
4 sites (10 ha)
(78 ha)
5 sites? 7 sites (31 ha)
15 sites
26 sites
30 sites
?
20 sites (70 ha)
19 sites 12 sites (45 ha)
11 sites
25 sites
41 sites (42 ha)
43 sites
53 sites (98 ha)
(45 ha)
48 sites
13 sites (64 ha)
(353 ha)
19 sites
7 sites (15 ha)
157 sites
(767 ha)
17 sites
19 sites
L
31 sites (38 ha)
2 sites
32 sites (74 ha)
few sites?
no
sites?
many sites
few
sites
?
2 sites
3 sites
LC 5
phase 4
3100
phase 3
EJ 0
phase 4
3000
phase 3
EJ I
13 sites (23 ha)
2800
2900
MB sites along Euphrates,
1 MB site in steppe area?
EJ II
K
LC/EB stone enclosures and cairns/tumuli in steppe area?, EB sites along Euphrates
2700
J
?
?
many sites
6 sites
phase 6
phase 6
?
EJ IIIa
1 site
2600
no data available
EJ IIIb
2500
40 sites (82 ha)
2400
relatively densely occupied?, including tell sites and walled sites
EJ IV
3 sites
2300
6 sites
EJ V
2200
phase 5
2100
phase 5
OJ I
21 sites (74 ha)
2000
13 sites (44 ha)
OJ II-III
5 sites (31 ha)
1700
1900
I
57 sites 43 sites
(100 ha) (82 ha)
MJ I
1800
H
45 sites
G
~25
sites?
F
38 sites
E
38 sites
38 sites (121 ha)
(101 ha)
D
27 sites 16 sites 12 sites
C
25 sites
B
27 sites (72 ha)
1600
A
5 sites
date BC Jezirah
no data available
Directly to the west of the Khabur, the area around the Jebel Abd al-Aziz as far west as a
north–south line crossing Ras al-‘Ain was surveyed as part of the Yale University Khabur Basin
Project (Fig. 5.1U).69 The area can be considered marginal from the viewpoint of agriculture,
which today is carried out with irrigation water supplied by diesel pumps (Hole 1997a: 44). The
early third millennium sites, that is those sites dating to EJ I–II, were predominantly small and may
have been only seasonally occupied. This settlement system changed dramatically during EJ III,
when large villages and walled towns up to 60 ha in size appeared. The number of sites also grew
considerably from EJ I–II to EJ III. The large EJ III settlements, typically Kranzhügel sites, were
located in areas where water was available, typically near karstic sinks and the tails of alluvial fans.
Even so, the carrying capacity of the site catchments could not support archaeologically inferred
population levels (Kouchoukos 1998: 393). Kouchoukos suggests that there is no evidence for a
settlement hierarchy that would support Wilkinson’s extraction model, and therefore concludes
that these large settlements probably relied on a broad-spectrum economy including agricultural,
pastoral, and wild resources (1998: 394). There is apparently no evidence for permanent settlement
in the region between 2500 bc, when the Kranzhügel settlement system ended, and the Neo-Assyrian
period.
Fig. 5.14: Uncorrected settlement trends in survey regions in northern Mesopotamia for the period 3100–1500 bc. Shading
indicates relative intensity of settlement. (A) Birecik-Euphrates Dam Survey (B-EDS). (B) Tishrin Dam surveys. (C) Tabqa Dam
surveys. (D) Land of Carchemish survey. (E) Westgazira survey. (F) Balikh-Euphrates uplands survey. (G) Middle Euphrates
survey. (H) Balikh Survey (BS). (I) Harran survey. (J) Jebel Bishri survey. (K) Wadi Hammar survey. (L) Upper Khabur Survey.
(M) Tell Beydar Survey (TBS). (N) Brak sustaining area survey. (O) 1988 Tell Brak Survey. (P) Leilan Regional Survey (LRS). (Q)
Tell Hamoukar Survey (THS). (R) Northeast Syria survey. (S) North Jazira Survey (NJS). (T) Lower and middle Khabur survey.
(U) Khabur Basin Project. (V) Wadi Ağiğ survey.
69 The survey was carried out in four seasons between 1988 and 1994. The survey was performed by car along systematic
transects that were planned on the basis of satellite images. Over 150 sites were recorded (but see Hole 1997b: 44, who
mentions 271 sites), of which 50 dated to the third millennium (Kouchoukos 1998: 365–7).
98
The last survey to be discussed here, that of the Wadi Ağiğ, aimed to document traces of permanent occupation in the region east of the lower Khabur (Fig. 5.1V), but west of the Syrian–Iraqi
border.70 Fourth millennium bc material was present at several sites, of which two were probably
permanently occupied. The material at other sites may have been transported there, or may have
reflected temporary occupations, for example for agricultural activities. Contrary to this settlement
pattern, the survey recovered virtually no third millennium material, and only forty-three early
second millennium sherds divided over seventeen sites. Given the absence of indications for permanent occupation, Bernbeck suggested that this material reflected temporary use of the sites by
mobile groups. The material from the Medieval Islamic period was similarly patterned and could
be interpreted along the same lines (Bernbeck 1993: 63).
5.4.2 The historical evidence
Texts from the Mari archive allow the reconstruction of the number of inhabitants for a number
of settlements within the core territories of the Mari kingdom under Zimri-Lim (Millet Albà
2004). A number of texts list villages in the districts of Mari, Terqa, and Saggarâtum and their
tribal affiliation, that is, whether they were considered to be Sim’alite or Yaminite. At least thirtynine Yaminite toponyms could be identified, and for twenty-nine of these, data on the number
of inhabitants are also available. These texts indicate that settlement sizes varied between as few
as 30 to as many as 675 inhabitants. For the district of Mari, seven Yaminite settlements are listed
with a total of 1500 occupants. For Terqa, nineteen Yaminite settlements are known; fourteen of
which had a combined population of 3875 persons. Finally, for the district of Saggarâtum, thirteen
Yaminite settlements are known, and of these eight had a combined population of 2225 persons
(Millet Albà 2004: 232). Thus, a total of 7600 persons lived in twenty-nine settlements, and at least
ten more Yaminite settlements are known.71 It is further mentioned that the total population of
these three districts may have been 40000 persons, although the origin of this figure is not explained. The district of Suhûm may have housed a further 10000 persons (Millet Albà 2004: 231).
Note, however, that the number of known toponyms exceeds the number of sites with a Middle
Bronze Age component (Millet Albà 2004: 225). This is perhaps to be expected in the alluvial environment of the Euphrates Valley.
5.5 Conclusions
The present chapter aimed to present a detailed reconstruction of demographic developments in
three different survey regions in northern Mesopotamia. Based on the criteria outlined in Section
5.1, the Birecik-Euphrates Dam Survey, the Balikh Survey (BS), and the North Jazira Survey (NJS)
were selected for analysis. These surveys, incidentally, also represent three different environmental
and cultural zones. The B-EDS region, located upstream of Carchemish along the Euphrates, was
probably always, or at least at the end of the third millennium, part of the cultural and economic
hinterland of Carchemish. The Balikh Valley can be considered a relatively autonomous area that
was, as far as can be gleaned from the historical sources, outside the permanent political control of
the large late third millennium political entities, such as Ebla, Akkad, and the Ur III state. The area
is also interesting because it represents a river valley that crosses the long-term border of dry agriculture. The NJS covers a region without major perennial streams, but with potentially sufficient
precipitation for dry agriculture. Today, the area is above the 400 mm isohyet, meaning that even
in dry years it is likely that dry agriculture will succeed. Politically, the area was certainly within the
70 In an area of 1150 km2, sites and sherd scatters were identified by asking local herders, who showed the sites to the
survey team. Large sites were divided into collection units based on site morphology, whereas no such distinction was
made for small sites. Sherds were unsystematically collected until an amount was reached that was deemed sufficient
for chronological and stylistic analysis (Bernbeck 1993: 19).
71 Average settlement size calculated from these numbers would be 262 persons, the median size is 195. In terms of settlement size, and assuming an on-site population density of 150–250 persons/ha, a settlement of 260 persons would
be roughly 1.0–1.7 ha.
99
influence sphere of Akkad and Ur III, but in the absence of local historical sources, it is difficult
to assess the impact of these states on the region. The exact implications of these differences will
be evaluated in Section 8.2. For the moment, the following observations can be made.
First, each region was characterized by a different population development. The B-EDS region
showed a very low overall population and very gradual population developments. The population
in the Balikh Valley grew toward the middle third millennium bc, which is in line with the general
trend toward urbanization in northern Mesopotamia as a whole. Average settlement size, however,
remained at the same level, indicating that the population growth was perhaps more a by-product
of urbanization elsewhere than a truly indigenous development. The population of the NJS region
grew steadily from the early third to the early second millennium bc. Although the coarse chronology may have masked some minor fluctuations, the overall trend is very clear.
Secondly, there appears to have been no clear correlation between population growth, and settlement system organization as reflected in the various diachronic rank-size plots. In the BS region,
there is a trend toward a convex site size distribution from the mid-third to the early second millennium, after which point convexity seems to have decreased again. From 2400 to 1800 bc, there
was a population decrease followed by an increase. Potentially, the settlement organization development could be in step with the population development albeit with a 100–200 year time-gap, but
this would need much more refinement of the settlement data in order to be tested.
Finally, there are various periods during which population growth exceeded historical natural
growth rates, and periods during which populations became smaller. There is some evidence that
population growth in one region might be balanced by population decline elsewhere. In fact, the
NJS region, which harboured the largest number of sites and the highest absolute population level,
demonstrated a consistent population growth that relatively closely approximated the historical
natural growth rates.
100
6 The development of pastoralism
6.1 Introduction
Section 3.5.2 argued that environmental stress is likely to result in the formation of social bonds
between groups practising different subsistence strategies. This chapter aims to review the evidence for pastoralism in northern Mesopotamia during the third and early second millennium
bc. The central question that will be addressed here is: were there specialized nomadic pastoralists in northern Mesopotamia during the third and early second millennium bc? This question is
important, because regional settlement abandonment has been regularly linked to the nomadization of society. This chapter first defines important concepts in the discussion on nomadism and
pastoralism (Section 6.2), and outlines the development and organization of the pastoral mode of
production up to the beginning of the third millennium bc (Section 6.3). Section 6.4 discusses the
zoological and historical evidence for pastoralism during the third and early second millennium bc,
and Section 6.5 the development of pastoralism during that period in terms of organization, specialization and mobility. Combined with the evidence presented in Chapter 7, this model will serve
as the basis for evaluating the hypothesis that environmental stress should lead to the emergence
of cooperative social mechanisms between pastoralists and agriculturalists (Section 8.3).
6.2 Defining pastoralism and nomadism
It is important to have a clear understanding of how the terms nomadism, semi-nomadism, trans
humance, and pastoralism should be understood, as they are often incorrectly used interchangeably, but actually represent different yet in many ways connected activities. Here, the definitions
provided by Cribb (1991: 18–20) will be followed. Nomadism refers to ‘the regular migration
of a community together with much of its productive base within a single ecological niche’. Seminomadism refers to a state of mobility on a scale with untied nomadism, that is, without fixed migration routes, at one end and a fully sedentary existence at the other. Especially transhumance is
a confusing term, as it is used to describe a variety of mobile subsistence activities. The definition
given by Cribb is as follows: ‘a form of livestock management making use of seasonal variations
in the availability of pasture’. Pastoralism is a mode of subsistence that exploits herd animals,
‘characteristically involving protection of the herd and systematic consumption of its renewable
products’. Herding can be understood as a more general term indicating how animals are managed,
but not how they are exploited. Thus, nomadism and semi-nomadism refer to communal mobility, whereas pastoralism, transhumance, and herding are principally subsistence activities. Cribb
further suggested that there seems to be a strong correlation between mobility and pastoralism on
the one hand, and sedentism and agriculture on the other hand, as far as can be gleaned from ethnohistorical case-studies. However, nomadic pastoralism and sedentary agriculture should not be
seen as mutually exclusive opposites, but as the extremes on a continuum, with most groups falling
somewhere in between (Cribb 1991: 15–17).72
72 The present discussion will be limited to sheep/goat pastoralism. The camel was not introduced into northern
Mesopotamia and the Levant until the late second millennium bc (Köhler-Rollefson 1996). Cattle was used for traction, and some cattle herding may have occurred (Stol 1995: 184), but it is likely that this was confined to river valleys
and other environments that could meet the higher water and food requirements of cattle, as compared to sheep/goat.
Redding suggested that cattle husbandry was probably more strongly associated with agricultural production than with
pure pastoral production (Redding 1993: 86).
101
In his study on the emergence of nomadic pastoralism in the Central Zagros, Abdi elaborated
on these terms by presenting a number of different pastoral production modes, based on degree
of mobility, division of labour, and social organization (2003: 400). He distinguished between village-based herding, transhumant pastoralism, semi-nomadic pastoralism, and nomadic pastoralism.
Village-based herding can be divided into proximate and distant variants. Proximate village-based
herding is organized along the principle that the herds return to the settlement on a daily basis, and
is often of only supplementary nature to agricultural practices. Also, it is not a full-time specialized
activity, but rather entrusted to younger members of the community who are not necessary for
the agricultural production process. The spatial and temporal organization of this type of herding is closely connected with the agricultural cycle, to make sure that the herds do not damage the
crops and to ensure the availability of fodder (Abdi 2003: 400–1). Distant village-based herding
involves taking the herds to pastures outside the area taken up by agriculture for extended periods
of time. However, these periods tend not to exceed several days, after which the herds return to
the village. This type of herding is normally carried out by non-specialist herdsmen, such as young
adults (Abdi 2003: 401). Under a transhumant pastoral production system, herding is delegated to
specialists and the herds are seasonally moved to pastures that are in their most productive season.
In mountainous areas, this means that the herds are taken to the mountains in summer and to the
lowlands in winter (Abdi 2003: 402–3). In a semi-nomadic pastoral economy, pastoral economic
activities are more important than agriculture, which is still practised by the whole group or part
of the group who specialize in it. ‘Seminomadic pastoralism is characterized by extensive herding
and a periodic change of pastures during the course of the entire or greater part of the year’ (Abdi
2003: 403). Finally, nomadic pastoralism involves no agriculture at all, not even in a supplementary
role. From this perspective, truly nomadic pastoralism occurred only in very specific environments
and regions, and semi-nomadic pastoralism with a stronger focus on pastoralism than on agriculture may have been more common (Cribb 1991: 16; Abdi 2003: 404). Other activities that nomads
engage in are trade, transport, or hiring themselves out as mercenaries or guides. In such cases, it
is probably better to speak of a nomadic multi-resource strategy, rather than pastoralism (DysonHudson and Dyson-Hudson 1980: 18; Abdi 2003: 404–5; Rosen 2003).
6.3 Animal husbandry and pastoralism in pre- and proto-history
The pastoralist mode of production is a fundamental part of the traditional Near Eastern subsistence production system. However, pastoralism as a subsistence strategy exploiting renewable
products is a relatively recent development and only arose after a long period of more generalized
herding strategies, in which animals were probably predominantly exploited for their meat. Sheep
and goat husbandry appeared during the early stages of the Pre-Pottery Neolithic B (8700–6800
cal bc) in the southern piedmont region of the Taurus Mountains, eventually leading to complete
domestication. The domestication of cattle probably started somewhat later, and domesticated pig
appeared around this time as well, although there is disputed evidence for a much earlier domestication date at the end of the Natufian period. From southeastern Turkey, the practice of animal
husbandry seems to have spread south at varying rates, presumably because animals had to adapt
to new environmental conditions, and cultural choice cannot be entirely ruled out either (Peters
et al. 1999).
There is some evidence for semi-nomadic exploitation of ovicaprine resources during the final
stages of the PPNB. The village site of El-Kowm 2 was probably occupied year-round, and the
occupants were involved in agriculture as well as animal husbandry. The nearby site of Qdeir was
much more ephemeral in terms of architecture, but agriculture was practised there as well, indicating that the inhabitants must have been at least seasonally present at the site (Stordeur 1993).
At ‘Ain Ghazal (Jordan), reliance on domestic goat increased steadily at the expense of hunted
animals from the late PPNB into the Yarmoukian period. It has been suggested that during the
PPNC at ‘Ain Ghazal, goat husbandry was practised by the site’s inhabitants, with herders possibly taking the flocks to seasonal pastures farther removed from the site to relieve pressure on
the catchment area. According to this scenario, the site catchment had been depleted during the
102
The development of pastoralism
Yarmoukian, resulting in the abandonment of the site by its sedentary agricultural population.
Herders could, however, continue to frequent the region for pasture, and this may have represented a first step towards a fully specialized domestic animal exploitation, and eventually nomadization (Köhler-Rollefson 1992). Although there is only indirect evidence for this scenario, recent
research indicates that a similar spatial separation between the settlement and its pastoral production may also have occurred at Çatalhöyük (Turkey) (Pearson et al. 2007). Pearson et al. conducted
an isotope analysis on bone collagen from sheep/goat bones from Pre Level XII D to Level VI
from Çatalhöyük East (7300–6400 cal bc). This study indicated that stable carbon (δ13C) and nitrogen (δ15N) isotope levels were relatively uniform in the early period, but that variability between
sampled bones increased towards the later periods. Because these isotopes are directly related
to the diet of the sampled animals, this change may reflect an increase in diet variability and the
consumption of a wide variety of plants from different ecological zones. Pearson et al. suggested
that this increased variability reflected ‘an increase in the significance of pastoralism with herders
moving separate flocks over more extensive territories around villages thus becoming more likely
to encounter multiple isotopically distinctive plant biomasses’ (2007: 2178). As at ‘Ain Ghazal, it is
possible that this shift in the spatial organization of pastoral production was linked to the need for
the population of Çatalhöyük to increase agricultural production, either through intensification, or
through an increase in the amount of land that was under production. Summarizing, the evidence
presented here suggests that some limited form of transhumant herding may have emerged at the
end of the PPNB or during the early Pottery Neolithic, although the degree of specialization and
the distances that were covered remain difficult to assess.
During the Late Neolithic in Syria (6800–5300 cal bc), ovicaprines were the dominant species
in faunal assemblages, although there were clear differences between larger, sedentary sites and
smaller, presumably short-lived sites where wild animals were also important. There is indirect
evidence for high mobility that can be associated with herding, again indicating the importance of
this subsistence strategy. At Late Neolithic Tell Sabi Abyad in the Balikh Valley, sealing practices
have been linked with the periodic absence of a part of the settlement’s population, which may
have been involved in herding. Early Chalcolithic subsistence in northern Mesopotamia is much
more poorly understood. During the Ubaid period in northern Mesopotamia, the importance of
ovicaprines increased to the detriment of cattle and pig. Ovicaprines were probably herded in
small flocks, and hunting remained important as well (Akkermans and Schwartz 2003).
During these early periods, ovicaprines must have been mainly exploited for their meat and
hides. Pastoralism as a subsistence strategy relying on renewable products could have only emerged
after the secondary products revolution. The secondary products revolution would have greatly
enhanced the value of sheep, that not only produced meat but also milk and wool. The selection
of wool-bearing and milk-producing qualities in animal breeds is, however, a long process that
would have yielded results only long after the initial domestication. For that reason, the secondary
products revolution has traditionally been placed somewhere in the fourth millennium (Sherratt
1981), although it is possible that Neolithic people used secondary products on a more modest and
unsystematic scale (Akkermans and Schwartz 2003: 74). In fact, the use of milk has recently been
attested for the middle seventh millennium bc in Anatolia and northern Mesopotamia (Evershed
et al. 2008).
The faunal assemblage of late fourth millennium Tell Qraya, south of where the Khabur joins
the Euphrates, indicated a ‘generalized, multipurpose form of herding’ including production for
dairy as well as meat, but probably not wool (Galvin 1987: 123). At other sites, however, wool was
certainly exploited during the late fourth millennium bc. Texts from Uruk indicated the existence
of wool-bearing sheep breeds, and archaeozoological evidence indicates that herding strategies favouring wool production were employed (McCorriston 1997: 521). Although it has been suggested
for northern Mesopotamia that specialization in sheep/goat pastoralism was primarily associated
with Uruk colony sites, it seems that this development also occurred as an indigenous development
before the occurrence of Uruk influence. Even though faunal material from many sites indicated
an increasing importance of sheep/goat, there were notable exceptions where wild resources remained important (Akkermans and Schwartz 2003: 206).
103
Abdi (2003) presented a model for the emergence of nomadic pastoralism in the Islamabad
Plain in the Central Zagros during the Middle/Late Chalcolithic. Based on archaeological survey
evidence, he suggested that from the Early Neolithic to the Middle Chalcolithic period, there was
a spectacular increase in the number and size of sedentary occupations, followed by a similarly
dramatic decline during the Late Chalcolithic. The Middle Chalcolithic also yielded the strongest
evidence for irrigation. Synchronously with the Late Chalcolithic sedentary settlement decline, the
number of temporary campsites increased, as did the distance between these campsites and the
permanent occupations. Excavations at selected permanent and temporary occupations revealed
significant differences relating to the activities that were carried out at each site. For example, the
faunal material from Chogha Gavaneh, a permanent occupation, indicated a diversified animal
economy, whereas the material from the temporary campsite of Tuwah Khoskeh indicated seasonal use focusing on sheep/goat animal husbandry. Pottery and flint tools were also indicative
of a rather limited array of activities, and large storage jars for example were absent. According
to Abdi, these developments highlighted an increasing distance between increasingly specialized
subsistence strategies. The evidence for intensive agriculture is contemporaneous with a settlement
peak and would have forced pastoralists to move their flocks over increasingly greater distances.
This development would in turn have provided to an incentive for pastoralists to specialize in herding because the movement and management of flocks demanded greater skill and knowledge.
Late fourth/early third millennium bc texts from Uruk attest to the highly developed and diversified pastoral economy that existed at that time. The texts list various breeds of sheep and
goat, as well as the products for which they were kept, including wool and various dairy products
(Szarzynska 2002). Based on the relative frequencies of attestations of sheep and goat, and the
fact that the signs for sheep breeds were much more diverse, it has been suggested that sheep represented the dominant animal in this economy (Szarzynska 2002: 7). 73 These texts may have been
herding contracts in which a flock was placed in the care of a shepherd for a period of one year.
The texts probably described the condition of the herd at the end of that year, so that the performance of the shepherd and the herd could be monitored (Green 1980: 15).
6.4 Pastoralism during the third and early second millennium
bc
Extensive evidence is available for pastoral production in northern Mesopotamia during the third
and early second millennium bc. The faunal material recovered from numerous excavations, textual evidence, and secondary evidence on settlement patterns and settlement types allow at least a
partial reconstruction of northern Mesopotamian pastoral production. This section will present an
overview of the available evidence, and aims to reconstruct changes through time in the organization of pastoral production.
Numerous sites in northern Mesopotamia have yielded archaeozoological data, allowing
inter-regional comparison of local production strategies, and elucidation of trends through time
(Fig. 6.1). Recent evaluations of pastoral production throughout northern Mesopotamia indicated
that there were clear differences between regions bordering the Taurus foothills and areas further north on the one hand, and the drier regions to the south on the other hand (Weber 1997;
Clason and Buitenhuis 1998; Zeder 1998). Both along the Euphrates and along the Khabur Rivers,
the importance of sheep/goat increased from north to south at the expense of other domesticated animals. The percentage of sheep/goat at fourth to early second millennium sites along the
Turkish Euphrates ranged between 40–70 % (Fig. 6.2A), whereas for sites in the Tabqa–Tishrin
Dam area, this percentage varied between 50–90 % (Fig. 6.2B). Similar percentages were observed at Hammam et-Turkman (Fig. 6.2C), the only site in the Balikh Valley to have yielded an
Early/Middle Bronze Age faunal assemblage (Buitenhuis forthcoming). At Tell Chuera (Fig. 6.2C),
70–90 % of the faunal assemblage consisted of sheep/goat, whereas wild animals were virtually
absent (Boessneck 1988; Vila 1995). At Umm el-Marra, on the other hand, sheep/goat were much
less dominant and accounted for only 40–60 % of the total assemblage in favour of gazelle and
73 Szarzynska (2002: 5) listed fourteen signs for sheep and five for goats, including such variants as ‘breeding adult male
sheep’ (udu utua), ‘fat-tailed sheep’ (gukkal), and ‘wool sheep’ (udu sig).
104
A
Gritille
Hayaz
Kurban
Hacinebi
Zeytinli Bahçe
D
Leilan
Arbid
C
Chuera
Hammam et-Turkman
B
El-Qitar
Hadidi
Sweyhat
Beydar
Brak
Kashkashok II-IV
Kuran
E
Shaqrah
Gudeda RadRaqa'i
Ziyadeh Atij
Bderi Umm Qseir
Mashnaqa
Ras et-Tell
Umm el-Marra Habuba Kabira
Selenkahiye
Al-Rawda
Terqa
Mari
2000+ m
1500-2000 m
1000-1500 m
500-1000 m
0-500 m
0 km
intermittent lake
lake
N
30
60
Fig. 6.1: Map of northern Mesopotamia indicating sites with faunal assemblages discussed in this study. The rectangles indicate
the site clusters used in Figs. 6.2 and 6.4. (A) Turkish Euphrates. (B) Syrian Upper Euphrates. (C) Balikh/Wadi Hammar. (D)
Upper Khabur. (E) Middle Khabur.
onager (Schwartz et al. 2000: table 2). Similarly, sheep/goat represented between 40–80 % of faunal assemblages from sites in the Upper Khabur (Fig. 6.2D). Along the Middle Khabur percentages
of sheep/goat ranged between 30–90 %, with a considerable increase in the percentage of sheep/
goat taking place since the late third millennium (Fig. 6.2E) (Zeder 1998). Both in the Euphrates
and Khabur assemblages, the importance of sheep/goat increased over time. 74
These differences between northern and southern sites in northern Mesopotamia may be partially explained by environmental conditions. However, other determinants may have played a role
as well. At various sites, differences in the faunal assemblage through time as well as between excavation areas have allowed the reconstruction of diversified local subsistence strategies.
Wattenmaker argued that at Kurban Höyük socioeconomic differences accounted for the variability in the faunal assemblages recovered from different contexts. She argued that households
of lower socioeconomic status relied more heavily on pigs, whereas elite households apparently
focused on sheep/goat. The sheep/goat assemblage from period IV.2/3 (2400–2200 bc) suggested that herding was an activity that was carried out by individual households. The absence of
prime-aged sheep/goat indicated tribute payments to a higher-order settlement or elite. Both elite
and non-elite households were involved in sheep/goat herding, but elite households may have
been able to focus primarily on sheep/goat whereas non-elite households had to rear other species as well, that is, cattle and pigs, to reduce risks associated with specialized subsistence strate74 The number of identified specimens (NISP) is used as indicator of species distribution, rather than the minimum
number of individuals (MNI) (see Renfrew and Bahn 1996 (1991): 272–3, for the potential biases resulting from this
choice). It should further be noted that only those data were used where bones could be identified to species level. If
animal remains could only be assigned to a particular size, or could not be determined at all, these data were excluded
from the present study.
105
40
5th mill.
4th mill.
3rd mill.
2nd mill.
80
60
40
Kashkashok III
Kuran
Kuran
Kuran
Brak
Kuran
Kashkashok II
Leilan
Beydar (ED III)
Beydar (Early Akkadian)
Arbid VI (EJ V)
Kashkashok IV
Arbid V (OJ II)
Arbid V late (OJ III)
Beydar (Mitanni)
20
0
100
Hacinebi (A)
Hacinebi (B)
Zeytinlibahçe Höyük (Late Uruk)
Hayaz Höyük (Chalcolithic)
Hayaz Höyük (EB)
Zeytinlibahçe Höyük (EBA IA)
Zeytinlibahçe Höyük (EBA IB)
Kurban Höyük (IV.4)
Gritille
Kurban Höyük (III)
Zeytinlibahçe Höyük (MBA)
Hayaz Höyük (MB/LB)
20
sheep/goat
other domestic animals
2nd mill.
80
60
40
20
0
5th mill.
E
3rd mill.
4th mill.
C
4th mill.
3rd mill.
3rd mill.
2nd mill.
100
80
60
40
20
0
Hammam et-Turkman (LC)
Hammam et-Turkman (EB II/III)
Chuera (Steinbau 1)
Hammam et-Turkman (EB III)
Chuera (Kleiner Antentempel)
Hammam et-Turkman (MB)
Hammam et-Turkman (LB)
60
100
percentage of total assemblage (NISP)
80
0
4th mill.
B
Habuba Kabira South
Tell Habuba Kabira (EB)
Sweyhat VI
Sweyhat VI/V
Sweyhat V
Selenkahiye
Sweyhat V/IV
Sweyhat IV
Tell Habuba Kabira (MB)
El-Qitar
Hadidi
2nd mill.
2nd mill.
80
60
40
20
0
Ziyadeh
Mashnaqa
Ziyadeh
Mashnaqa
Mashnaqa
Umm Qseir
Ziyadeh
Atij
Raqa'i
Gudeda
Mashnaqa
Rad Shaqrah 1-4 (EJ III)
Bderi
Rad Shaqrah 5-8 (EJ IV)
Bderi
100
D
3rd mill.
100
4th mill.
A
domestic/wild animals
wild animals
Fig. 6.2: Presence of sheep/goat, other domestic, domestic/wild, and wild animals at fifth to second millennium bc sites in
northern Mesopotamia (after Clason and Buitenhuis 1978; Stein 1987; Wattenmaker 1987; Boessneck 1988; Buitenhuis 1988;
von den Driesch 1993; Vila 1995; Stein et al. 1996; Rothman et al. 1998; Zeder 1998; van Neer and Decupere 2000; IJzereef
2001; Frangipane et al. 2004; Koliński and Piątkowska-Małecka 2008; Buitenhuis forthcoming). Note that in some cases, the
class domestic/wild could not be distinguished on the basis of the available data. (A) Turkish Euphrates. (B) Syrian Upper
Euphrates. (C) Balikh/Wadi Hammar. (D) Upper Khabur. (E) Middle Khabur.
gies. A shift in the slaughter-pattern towards younger animals probably occurred during period
IV.1 (2200–2100 bc), as did an increase in the importance of sheep/goat in the total assemblage,
suggesting that sheep/goat herding may have become a specialized activity carried out by fulltime herders, from whom the sedentary households at Kurban procured their animal products.
During period III (2100–2000 bc), household production seems to have been the rule again, as
evidenced by data on age of death, and the higher diversity of species recorded in the assemblage
(Wattenmaker 1987, 1998).
Evidence for a dual pastoral economy consisting of generalized pastoral production alongside specialized pastoral production came from Tell es-Sweyhat. Wilkinson (2004: 168) noted the
marked differences between analyses by Buitenhuis of faunal material from the excavations by
Holland, and by Weber of material from the more recent excavations by Zettler. Buitenhuis’ analysis of mid–late third millennium material (1988) indicated that ovicaprines made up 50–90 %
106
of the total assemblage, with especially high percentages recorded for the early periods. Weber’s
analysis of the material from nearby and contemporary Hajji Ibrahim demonstrated the existence
of a much more diversified animal economy, with greater percentages of pig and cattle (1997).
According to Wilkinson, this difference probably reflected the existence of two subsistence strategies. The wide range of animals identified at Hajji Ibrahim was incorporated in a village-based
production system, in which agriculture was combined with small-scale pastoral production. The
higher sheep/goat percentages of Tell es-Sweyhat resulted from a more specialized pastoralism,
involving seasonal migration or fully nomadic movement to pasture areas beyond the immediate
catchment area of Tell es-Sweyhat (Wilkinson 2004: 169).
The agro-pastoral economy of Tell Beydar during the EJ IIIb period is relatively well-known
due to the recovery of numerous texts from the Official Block and Chantier B, dealing with various aspects of the centralized institutional economy (Sallaberger 1996, 2004b). Although inconsistencies and uncertainties remain, these texts allow the following reconstruction of institutionalized pastoral production. All aspects of agro-pastoral production were overseen by five officials,
who each headed a public household. These households consisted of workmen who were assigned
various tasks, and who received monthly rations of grain. The largest group of workmen was involved in agricultural production and designated by lu2-giš-DU apin. The second-largest group
consisted of the ba-ri2 udu, probably to be translated as ‘the one overseeing the sheep’, and usually
associated with some form of sheep/goat husbandry. There were between 19 and 28 ba-ri2 udu associated with each household overseer, indicating that there may have been 100–40 ba-ri2 udu at the
same time. However, the exact function of the ba-ri2 udu is uncertain. If a sealing with the name
Ma-ti belonged to a ba-ri2 udu with the same name, it is possible that ba-ri2 udu was not a shepherd,
but some higher official. This is not necessarily the case, since at least three persons that go by the
name of Ma-ti are known from the Beydar texts. Furthermore, it is possible that the term ba-ri2 udu
indicated not only shepherd in the strictest sense, but was more widely applied to a ‘social group
living on animal husbandry’. These narrow and wide uses of the designation ba-ri2 udu could partly
explain why there are only twenty-eight ba-ri2 udu known by name to have been actually engaged
in herding activities (Sallaberger 2004b: 18).75
Several texts described the size and composition of herds of sheep or goat belonging to the
central institution of Beydar. Sheep and goat were herded in separate flocks. Only two out of
twenty-eight ba-ri2 udu involved in herding were associated with both sheep and goat, whereas sixteen were exclusively associated with sheep, and ten with goat. Each flock was entrusted to a ba-ri2
udu, who probably had to take care of the flock for the duration of one year. These texts all dated
to the Month of the Sun-God, which probably fell somewhere in spring-time. It is likely, therefore,
that the sheep were returned to Beydar during spring to be counted and plucked. The average size
of sheep herds was 210 animals, whereas goat herds consisted of 300 animals. The sheep were
designated as being destined ‘for plucking’ (ur). Sheep herds consisted of relatively large numbers
of female sheep, given the fact that rams gave more wool and would be expected to have been the
dominant sex if wool-production was maximized. Based on the available evidence, it has been suggested that the central institution of Beydar may have administered 7400 sheep and goat, divided
over thirty flocks (Sallaberger 2004b: 20).
Finally, the Beydar texts also recorded the rationing of grain to sheep (Sallaberger and Talon
1996: texts 7, 17). These sheep were described as ‘grain-fed sheep’ (udu niga) and udu nita HAR
(male … sheep) and were apparently different from the sheep that were designated by ur. The
number of grain-fattened sheep was much lower than the number of sheep that were herded in
flocks. Sallaberger suggested that grain-fattened sheep and grazing sheep were usually kept separate in Mesopotamia. Grain-fattened sheep may have been permanently kept at the site, whereas
grazing sheep were taken out to pasture areas (Sallaberger 2004b: 21). This division could reflect
the situation at Tell es-Sweyhat, where the faunal assemblage also indicated two different pastoral
production strategies.
75 Usually, shepherds are designated by the term sipa, but this word occurred only once in the Beydar tablets (Sallaberger
2004b: 18).
107
The faunal assemblages from the Middle Khabur sites clearly indicate that these settlements
were involved in specialized pastoral production during the early third millennium. It has been
observed that many early third millennium (EJ I–II) sites along the Middle Khabur were relatively
small, that is, below 1 ha in size, and had facilities capable of storing much more grain than would
be needed by the hypothesized populations. These sites clustered around the modern 250 mm
isohyet, and included Atij, Kerma, Raqa’i, Ziyadeh, and possibly several others (Hole 1999: 274).
There has been considerable debate on the function of these storage facilities. Schwartz (1994)
suggested that these sites provided agricultural surplus production for Mari, but this explanation
failed to account for the relatively marginal location of these sites. As an elaboration of this argument, it was suggested that these sites were way stations for grain or trade transports coming
from the north and going south along the Khabur (Margueron 1991: 99). The main argument for
this thesis was that the Khabur may have been navigable in the third millennium, and that anchor
stones were found at Tell Atij (Fortin 2000: 122). However, the evidence for higher water levels of
the Khabur during the third millennium is inconclusive, and Hole has calculated that the surplus
that could be produced by the population of Raqa’i was much smaller than suggested by Schwartz
(Hole 1999: 277–8). It is more likely, therefore, that any surplus was used locally, that some storage volume was not devoted to grain, but to other products such as wool, or that not all storage
facilities were in use at the same time. According to Hole, the Middle Khabur storage facilities
were used by small-scale communities that were involved in both agriculture and pastoralism, with
agriculture probably being of secondary importance to provide fodder for flocks (1999: 280). 76
Many researchers have associated the Kranzhügel phenomenon with pastoral activity. Most
Kranzhügel are found in a broad zone extending from the Wadi Hammar region in the northwest
to the Jebel Abd al-Aziz in the southeast (Fig. 6.3). However, in the absence of excavations of
Kranzhügel, with the exceptions of Beydar and Chuera, many of these associations remain rather
hypothetical (e.g. van Liere 1963: 114; Lyonnet 1998; Meijer 2000). Recently, Kouchoukos (1998)
carried out an extensive analysis of the environment of the Kranzhügel and contemporary settlements around the Jebel Abd al-Aziz, evaluating their agricultural potential. He first reconstructed
potential population levels. The excavations and geophysical research at sites like Chuera and
Beydar have shown that both the inner and outer towns of Kranzhügel could be relatively densely built up, suggesting that Kranzhügel may have housed considerable populations (Meyer 2006).
Oppenheim’s observations on the presence of significant stone architecture at other Kranzhügel
suggest that their layout probably resembled those of Chuera and Beydar (Moortgat-Correns 1972:
26ff.). Kouchoukos then compared these figures with the regional carrying capacity for barley agriculture. Based on the conclusion that the catchment areas of EJ III settlements could not support
the reconstructed populations, he argued that the inhabitants of these settlements must have relied
heavily on pastoral production in order to supplement their diet.
Indirect evidence suggests a similar strong reliance on pastoralism at the site of Al-Rawda
(2400–2100/2000 bc). Geophysical research at the site has revealed a dense urban layout within
the city wall. During the later occupation phases of the site, houses were also built outside the
city wall. Thus, as with the Kranzhügel further north, Al-Rawda may have housed a considerable
urban population (Castel 2007: 173). The site is located well below the 250 mm isohyet, suggesting that agricultural production without irrigation was unreliable. Al-Rawda is however favourably
located on a fayda, or natural depression, taking full advantage of any precipitation and runoff.
This advantage was further enhanced by the construction of hydraulic installations in the vicinity
of Al-Rawda (Barge and Moulin 2008). Analysis of botanical remains from the site has shown that
agriculture was an important subsistence strategy. Next to barley, crops such as vines and olives
were also cultivated, probably requiring some form of irrigation or water management (Herveux
2004). Despite the fact that agriculture was important for the inhabitants of Al-Rawda, pastoralism
may also have been an essential subsistence strategy. The presence of numerous large desert-kites
76 Recent research indicated that third millennium inhabitants of Raqa’i, or at least persons that were buried there,
consumed relatively small amounts of barley and wheat. Analysis of dental calculus indicated that plant microfossils
identifiable as wheat or barley made up between 7–16 % of the total amount of starches from grass seeds. The staple
plant food could unfortunately not be identified (Henry and Piperno 2008).
108
Khanzir
Chuera
Banat
Beydar
Abu Sheikha
Mugair
Mabtuh Gharbi
Mabtuh Sharqi
Raqa'i
Ras et-Tell
Mu'azzar
Malhat ed-Deru
Al-Rawda
Mari?
2000+ m
Kranzhügel
1500-2000 m
other circular site
1000-1500 m
other site
500-1000 m
0-500 m
intermittent lake
0 km 30
60
lake
Fig. 6.3: Map of northern Mesopotamia indicating the zone where Kranzhügel are found (based on Moortgat-Correns 1972: map
2; Lyonnet 1998: fig. 4).
in the wide surroundings of Al-Rawda makes a convincing case for the existence of a large-scale
pastoral economy in this area, in which Al-Rawda may have played an important role (Barge and
Moulin 2008: 20–1).77
For the Ur III period, there are numerous texts dealing with agricultural and pastoral production from the core area of the Ur III state. Although the conditions there were dissimilar to those
of northern Mesopotamia, some comments here may serve to illustrate the development and importance of pastoral production in wider Mesopotamia, especially because textual and zoological
sources from northern Mesopotamia are relatively rare for the late third–early second millennium
bc. The Drehem texts mainly documented the centralized state-economy, although a significant
private economy probably existed as well. Bureaucratic control of pastoral production is particularly well-known thanks to the Drehem archives. Drehem functioned as a redistribution centre,
mainly for animal products. Animals, including large numbers of sheep/goat were brought in and
processed at or around the site, and secondary products, as well as the animals themselves, were redistributed to various officials. Sheep/goat were brought in during the entire year, but it seems that
the number of incoming animals increased during the fall and winter months. During this period,
175 animals were delivered daily, whereas during summer and spring, 10 animals were brought in
per day. This discrepancy between winter and summer figures could reflect the presence of herders
around Drehem who were normally away on seasonal movements. Indirect evidence from Umma,
however, suggests that the distances covered by herders may have been relatively small. Herders
77 On the proposed function of so-called desert kites in a pastoral economy, rather than as structures facilitating (gazelle)
hunting, see Échallier and Braemer 1995.
109
were contracted to care for flocks of sheep/goat and deliver production to the central administration. It is possible that herding was not carried out by a single herder, but that he was assisted by
several others or that herding was carried out on a family basis (Zeder 1994: 179; Adams 2006).78
The evidence for specialized pastoral production in northern Mesopotamia is much less clear
during the early second millennium. The number of excavations that yielded faunal data is small,
but it seems that the percentage of ovicaprines in all regions analysed here (Figs. 6.2, 6.4) is slightly
larger than during the third millennium. Evidence from Terqa for the period 1750–1500 bc suggests that sheep and goat dominated the faunal assemblage at 37 % and 40 %, respectively. Cattle
were much less dominant at 11 % and pig was virtually absent. Sheep seem to have been kept for
wool production and some minor meat production as well, whereas goat may have been primarily
kept for meat production. However, the degree to which wool and meat production were specialized seems to have been limited (Galvin 1987).79
The Mari texts allow a fragmented reconstruction of the organization of the early second millennium pastoral economy (e.g. Kupper 1957; Luke 1965; Rowton 1974; Matthews 1978; Buccellati
1990; Streck 2002). Many details are not recorded, and especially practical issues such as breeding
practices, or the composition of herds, remain completely unknown. On the other hand, the texts
do provide detailed information on many socio-cultural aspects, including migration patterns,
pastures, and the social and economic position of pastoralists. Nevertheless, other social aspects
are not covered at all, and can only be reconstructed after careful analysis. Any reconstruction of
pastoralism during the early second millennium bc must therefore take into account the rather onesided nature of the texts that form the basis for the reconstruction.
The central concept in Mari pastoralism was the nawûm. The term nawûm had several related
meanings, summarized as ‘the total pastoral group and its surroundings’ (Matthews 1978: 59).
More specifically, nawûm could be used to designate the encampments of hana pastoralists, groups
of hana pastoralists, sheep/goat flocks, and pasture areas (Matthews 1978: 60–2; Anbar 1991: 161).
Where nawûm was used to designate a pastoral group, it was probably used on different scales, from
relatively small numbers to groups of thousands of people. The term nawûm, then, referred to various aspects of a type of pastoralism that was carried out outside settlements (see Section 6.5.2).
The terms hasiratum and rubsatim, respectively translated by Matthews as ‘camps’ and ‘pens’,
(Matthews 1978: 52), possibly referred to small-scale pastoral production in settlements. These
terms perhaps reflected the continued dichotomy between small-scale pastoral production in settlements, and large-scale specialized pastoral production in areas beyond the agricultural catchments of these settlements.
Several Mari texts describe migration routes and pasture areas, thus providing insight into
the degree of mobility practised by pastoralists. The pastoralists’ movements were monitored by
the state when they were within the core territory of the Mari kingdom, that is, Ah Purattim or
the ‘Banks of the Euphrates’. There are texts describing herders with their flocks crossing the
Euphrates, and describing the grazing of flocks on agricultural fields. Other texts described how
sheep/goat flocks were moved in response to (un)favourable grazing conditions (Matthews 1978:
55, ARM II 102, 9–16): ‘The men who were living at Sagarātum have turned their faces toward
the Upper Country only, and with their flocks which were grazing on laskim (?), they have set out
for the Upper Country. When he (the informant sent to investigate the situation by Yaqim-Addu)
questioned them, they answered in this manner: “There is no pasture (here). Therefore, we are setting out for the Upper Country.”’ Another text described how sheep/goat flocks of Yasmah-Addu
were moved to the territory of Qatna, because the pasture was more abundant there (Matthews
1978: 54). It has recently been proposed that the term nighum may have served to designate the
hana summer pastures, whereby different areas in northern Mesopotamia were assigned to either
the Sim’alites or the Yaminites. More specifically, Ida-Marash, in the Upper Khabur, was described
as Sim’alite territory, whereas Yamhad and Qatna were Yaminite pasturages (Durand 2004a: 391–2,
2004b: 118–21). Thus, it is likely that seasonal migration patterns existed, suggesting that some
78 According to Russell (1988: 83, 85), flocks of up to 400 sheep/goat can be herded by a single herder, although assistance of another full-time herder and a part-time herder is needed during the breeding season.
79 Galvin used MNI instead of NISP, and the figures for Terqa are, therefore, not included in Figs. 6.2 and 6.4.
110
archaeozoology
administrative texts
literary texts
+
-
-
mid-third millennium
+
+/-
-
late third millennium
+/-
+
-
-
-
+
early third millennium
early second millennium
Table 6.1: Comparison of available evidence for third and early second millennium bc pastoral subsistence strategies (+: strong
evidence; +/-: some evidence; -: very little/no evidence).
sort of transhumance was practised over relatively large distances in comparison to earlier periods.
As part of these seasonal migrations, flocks could graze on agricultural fields, presumably after the
harvest when grain stubble provided abundant food for sheep/goat.
Looking more closely at the data presented in this section, it becomes clear that there is a disparity in the nature of the evidence of the third millennium bc, as opposed to that of the early
second millennium bc (Table 6.1). It is possible that this discrepancy creates differences in the observed organization of the pastoral economy. This possibility is particularly true for the zoological
and textual evidence. The number of second millennium sites with faunal assemblages is relatively
small compared to the number of third millennium sites. This discrepancy and the fact that there
were relatively large inter-site differences between third millennium bc faunal assemblages suggest that the confidence with which second millennium regional animal husbandry patterns can be
reconstructed is relatively low. In the textual evidence, there is a shift from administrative toward
literary texts as the main source on pastoral subsistence strategies at the third–second millennium
transition. These disparities must be carefully considered in the evaluation of third–second millennium pastoralism that is the subject of the following section.
6.5 The emergence of specialized pastoralism in northern Mesopotamia
The present review has shown that by the end of the fourth millennium bc, a fully developed pastoral production system had emerged. At least from that time onwards, a large variety of breeds
was exploited for meat as well as for a full array of secondary products. Before the third millennium, there had been periods when economic and environmental circumstances necessitated a
removal of pastoral production from the direct environment of a settlement. Thus, at ‘Ain Ghazal
and Çatalhöyük, environmental degradation, growing population, or a combination of both possibly led to the development of some form of distant village-based herding, and the same development has been reconstructed for the Islamabad Plain during the Chalcolithic period. It seems,
then, that spatial separation between agricultural and pastoral production emerged at various places and various points in time as an adaptive strategy to the demands of agricultural intensification
and/or population pressure. The present section discusses the development of specialized pastoralism during the third and early second millennium bc.
6.5.1 Specialized pastoralism during the third millennium
bc
A similar scenario of increasing spatial separation of specialized pastoralism and agriculture can
be postulated for the Upper and Middle Khabur, and possibly to a lesser extent for the Syrian
Upper Euphrates during the mid-third millennium. Zeder has suggested that the faunal patterns
from third millennium sites in the Upper and Middle Khabur not only reflected environmental
differences, but also differences in the dominant pastoral production strategy (1998). The faunal
assemblages from sites along the Middle Khabur such as Raqa’i and Ziyadeh were characterized by
ever greater percentages of sheep/goat, at the expense of other domesticates such as pig and cattle, and wild animals. At the same time, the importance of sheep/goat at northern sites decreased,
with other domesticates becoming more important. According to Zeder, these developments resulted from the emergence of large urban settlements during the early to mid third millennium bc.
She suggested that the local economy of the Middle Khabur, characterized by exchange between
sedentary agro-pastoral communities along the Khabur, and more mobile fully pastoralist groups,
111
was eventually incorporated into the urban economy of the Upper Khabur. In fact, the highly
specialized Middle Khabur economy may have played a key role in the formation of urban society
further to the north (Zeder 1998: 65).
However, Zeder did not fully explain why specialized pastoral production emerged along
the Middle Khabur, and not in areas located closer to the urban centres of the Upper Khabur.
Wilkinson (2003) suggested that there may not have been sufficient land in the Upper Khabur to
sustain large flocks of sheep/goat. He used Corona satellite images to identify tells, and plotted
sustaining areas of 3 and 5 km around each of them. Even if agriculture was highly productive,
resulting in relatively small cultivated areas around each site, only a limited amount of land would
have been available as pasture (Wilkinson 2003: 121–2). Thus, urbanization in the Upper Khabur
may have provided both the incentive for specialized pastoral production, as well as the necessity for relocating (parts of) this production into the less densely inhabited regions of the western
Upper Khabur and the Middle Khabur. The survey evidence for occupation in the area between
the Euphrates and the Balikh, and around Tell Chuera, also points to increased activity during the
mid-third millennium (Fig. 5.14). This could indicate that a similar process may have been operating in the western Jezirah as well, although perhaps on a less massive scale than in the Khabur region. In fact, the evidence from Tell es-Sweyhat suggests that specialized pastoral and generalized
agro-pastoral production strategies may have been carried out in relatively close proximity. One
explanation may be that pasture lands unsuitable for agriculture could be found in close proximity
to Tell es-Sweyhat, that is, outside the Euphrates Valley, and that there was less incentive to keep
the two economies strictly separated. Alternatively, the assemblage at Tell es-Sweyhat reflects the
fact that urban communities consumed pastoral products produced mainly under a specialized system, which was not necessarily carried out near the site.
Hole (1999) suggested that by the mid-third millennium (EJ III), the small-scale riverine agropastoral communities along the Middle Khabur were gradually replaced by larger and more specialized pastoral communities around the Jebel Abd al-Aziz. He argued that the sequence of small
storage sites along the Khabur and the slightly later Kranzhügel and other large sites near the Jebel
Abd al-Aziz reflect a development whereby pastoralists became increasingly specialized and relocated their activities from the Middle Khabur toward the west. This development would have
occurred at a time when urban settlements in the north such as Mozan, Leilan, and Tell al-Hawa
were flourishing, and demand for pastoral products would be high. Thus, although it is perhaps
premature to point to urbanism as a causal mechanism in the emergence of specialized pastoralism in this region, the synchronous development is at least indicative of such a correlation. It is
likely that this development reflected a monetization of the pastoral and agricultural economies,
as is possibly indicated by texts from Beydar that record the hiring of herders to take flocks to the
pasture areas.
No third millennium site around the Jebel Abd al-Aziz has been excavated to test the importance of pastoralism and its degree of specialization. However, there is circumstantial evidence
to suggest that the region was visited by mobile communities during the mid-third millennium.
The site of Ras et-Tell on top of the Jebelet el-Bēdā, a small hilly outcrop southwest of the Jebel
Abd al-Aziz, may have functioned as a central place for surrounding (semi-)nomadic communities. Ras et-Tell was located at 775 m above sea-level, providing excellent views in most directions. Several basalt sculpture fragments were found on the slopes surrounding Ras et-Tell, and
it is likely that the sculptures, from which these fragments originated, originally stood on the top
of the hill. There must have been at least four sculptures at the site, all of which showed men
in fleece skirts. At and around Ras et-Tell, some 20 stone cist graves were found. No grave gifts
were associated with these graves, and they were marked on the surface by stone circles filled with
smaller pebbles (Moortgat-Correns 1972). This site has usually been interpreted as a victory monument based on the sculptures’ stylistic links with southern Mesopotamia (Moortgat-Correns 1972:
21–4). However, it was recently proposed that the site functioned as a regional burial and ancestor cult centre (Meyer 1997). According to Meyer, the grave groups may have reflected separate
burials by clans or lineages of (semi-)nomadic groups living in the area of the Jebel Abd al-Aziz.
The apparent absence of settlements in the immediate vicinity of Ras et-Tell might support this
112
4th mill.
3rd mill.
2nd mill.
80
60
40
0
Kashkashok III
Kuran
Kuran
Kuran
Brak
Kuran
Kashkashok II
Leilan
Beydar (ED III)
Beydar (Early Akkadian)
Arbid VI (EJ V)
Kashkashok IV
Arbid V (OJ II)
Arbid V late (OJ III)
Beydar (Mitanni)
20
60
40
20
5th mill.
E
percentage of total domestic (NISP)
100
80
0
4th mill.
C
4th mill.
3rd mill.
3rd mill.
2nd mill.
100
80
60
40
20
0
Hammam et-Turkman (LC)
Hammam et-Turkman (EB II/III)
Chuera (Steinbau 1)
Hammam et-Turkman (EB III)
Chuera (Kleiner Antentempel)
Hammam et-Turkman (MB)
Hammam et-Turkman (LB)
20
2nd mill.
40
3rd mill.
100
Habuba Kabira South
Tell Habuba Kabira (EB)
Sweyhat VI
Sweyhat VI/V
Sweyhat V
Selenkahiye
Sweyhat V/IV
Sweyhat IV
Tell Habuba Kabira (MB)
El-Qitar
Hadidi
60
0
4th mill.
B
Hacinebi (A)
Hacinebi (B)
Zeytinlibahçe Höyük (Late Uruk)
Hayaz Höyük (Chalcolithic)
Hayaz Höyük (EB)
Zeytinlibahçe Höyük (EBA IA)
Zeytinlibahçe Höyük (EBA IB)
Gritille
Kurban Höyük (III)
Zeytinlibahçe Höyük (MBA)
Hayaz Höyük (MB/LB)
2nd mill.
80
5th mill.
D
3rd mill.
100
2nd mill.
100
80
60
40
20
0
Ziyadeh
Mashnaqa
Ziyadeh
Mashnaqa
Mashnaqa
Umm Qseir
Ziyadeh
Atij
Raqa'i
Gudeda
Mashnaqa
Rad Shaqrah 1-4 (EJ III)
Bderi
Rad Shaqrah 5-8 (EJ IV)
Bderi
4th mill.
A
Fig. 6.4: Proportion of sheep/goat in the total assemblage of domestic animals at fifth to second millennium bc sites in northern
Mesopotamia. Sites are chronologically ordered. (A) Turkish Euphrates. (B) Syrian Upper Euphrates. (C) Balikh/Wadi Hammar.
(D) Upper Khabur. (E) Middle Khabur.
interpretation. Although the date of the monumental sculptures remains problematic, it is likely
that they date to EJ II–III, making them contemporaneous with the emergence of Kranzhügel in
the Jebel Abd al-Aziz region.80
The so-called White Monument of Tell Banat North may have represented a similar attestation
of tribal presence in northern Mesopotamia. This striking landmark was located along the Upper
Syrian Euphrates and consisted of a conical mound that was reconstructed and modified several
times during its lifespan from 2600 to 2300 bc. The White Monument was part of a larger complex
that at one time included not only the monument itself, but also a complex of partly public, partly
private buildings at Tell Banat. The activities carried out at the Tell Banat complex changed repeatedly, and the White Monument was constructed in three stages, indicating shifts in the function of
the complex as a whole. Both the building complex and the White Monument must have played
80 Ristvet argued that the cultic complex of Jebelet el-Bēdā, as well as the monumental sites of Banat and Hazna 1
reflected a ‘landscape of pilgrimage’ (2005: 105) in which city-based kings travelled to the countryside to legitimize
their authority and help ‘construct a new form of political landscape–that of kingdoms, not isolated villages’ (2005:
112).
113
a key role in funerary traditions, as the activity area of Tell Banat contained several stone-lined
tombs, and the White Monument yielded burials as well. Based on these considerations, Porter has
demonstrated that, throughout its entire history, the Banat complex functioned as a focal point for
pastoral communities enacting rituals and burying their dead (Porter 2002). 81
The relocation of specialized pastoral production to non-urban areas is furthermore suggested
by the importance of the sheep/goat faunal assemblage of sites in the Khabur region. At sites in
the northern Khabur area, the proportion of sheep/goat decreased in favour of other domestic
animals, including pig, suggesting that production was carried out at a small-scale household level.
At the Middle Khabur sites, however, the proportion of sheep/goat not only increased in the total
faunal assemblage, but also at the expense of other domestic animals. The contemporaneity and
inverse correlation of these developments fits nicely in the scenario presented above (Fig. 6.4D,
E). This development seems to be absent along the Euphrates and the Balikh, again suggesting
that the spatial separation between pastoral and agricultural production may have been less severe
(Fig. 6.4A–C).
Thus, by the mid to late third millennium bc a fully specialized pastoral economy had emerged,
centring on settlements located in areas where no large-scale occupation had existed previously,
and spatially separated from the urban core-areas. Although pastoralism was the main subsistence
strategy, agriculture was carried out on a small scale as well. Kouchoukos (1998) has shown that
the large settlements to the west of the Middle Khabur were located in areas that received runoff
from the Jebel Abd al-Aziz. It seems likely, then, that these communities practised some form of
distant village-based herding, whereby herders took the sheep/goat flocks to pastures around the
settlements, whereas other community members tended the small-scale agricultural fields, taking full advantage of rainfall, runoff, and soil moisture. The large sedentary communities in the
river valleys of the Euphrates, Khabur, and Balikh provided the necessary outlets for the pastoral
production of the pastoral communities in the upland areas, notably textiles (McCorriston 1997;
Kouchoukos 1998).
A similar scenario has been envisaged for the site of Al-Rawda in western Syria. The site’s favourable location with respect to agricultural productivity has already been pointed out, as well
as the central function the site may have fulfilled in the region’s pastoral economy (Section 6.4).
The circular layout of Al-Rawda makes it likely that the site was a planned settlement constructed
within a short period of time. It has been suggested that Al-Rawda was founded by people from
Ebla, or that the site at least had close economic ties with that important centre to the west. If
so, Al-Rawda may have fulfilled a key role in pastoralism and textile production. Additionally, AlRawda was strategically located along the trade route between the urban centres of the Orontes
Valley, and those of the Euphrates Valley (Castel and Peltenburg 2007).
6.5.2 Specialized pastoralism during the early second millennium
bc
The transition from the third to the second millennium is much more poorly understood. There
are no texts for this period that shed light on the organization of the subsistence economy, making
reconstructions of this economy rather hypothetical. The few faunal assemblages dating to this
period suggest that little changed with respect to the mid-third millennium in terms of exploited
animals. Thus, a dual pastoral economy with a specialized and a small-scale component probably
continued to exist. By the time of the Mari archives, however, a shift in the organization of the
specialized pastoral economy must have occurred. This is suggested by the abandonment of sites
in the more arid upland zones, as indicated by the Wadi Hammar survey and the survey around the
Jebel Abd al-Aziz (Fig. 5.14), and by the appearance of a surface assemblage in the Wadi Ağiğ that
displays many similarities with more recent assemblages that can be associated with nomadic pastoralists (see Section 5.4.1). It seems, then, that at least in some regions the mid-third millennium
village-based herding system was replaced by (semi-)nomadic pastoralism.
81 Note that there are several mounds throughout northern Mesopotamia that, based on their size and shape, are hypothesized to contain structures similar to the White Monument of Tell Banat (Porter 2002: n. 31).
114
It has recently been suggested that the term hana or hanûm occurring in the Mari archives
should be translated as ‘tent-dweller’ and that it cannot be taken to represent a tribal or ethnic
identity similar to Sim’alite or Yaminite. According to Durand, the term hana derived from the
root hn’, ‘to live in a tent’. Because there are several texts describing Yaminites and Sim’alites as
hana, it is likely that hana was a designation for a particular subgroup of Yaminites or Sim’alites,
rather than a tribal or ethnic group (Durand 1998: 417–18; Streck 2002: 175; Fleming 2004: 47–9).
Previous studies interpreted hana or Hanean as a tribal or ethnic group, partly on the grounds that
the designation occurred very regularly in the Mari texts. It is, however, plausible that the term
hana without ethnic or tribal determinative was very specifically used to designate Sim’alite tentdwellers. This may have been the case because the Lim-dynasty was of Sim’alite affiliation and
writers may not have felt the need to mention this affiliation because it so obviously dominated
the political scene of Mari.82
The equation of hana with mobile pastoralist is further strengthened by the fact that in some
cases hana was opposed to the town (ālum) or townspeople (Fleming 2004: 48, n. 63). Thus, in
ARM 8 11, a distinction is made between members of the tribal elite living in the town of Appan,
and those living with the nawûm. Another text from the reign of Yahdun-Lim listed soldiers according to their descent. This list clearly distinguished between soldiers classified as lu 2-meš hana
and dumu-meš mâtim, that is, hana people and sons of the land, respectively. The dumu-meš mâtim
were furthermore identified by district and town, making clear that this list distinguished between
(semi-)nomadic hana and sedentary townspeople (Charpin 2004). The mobility of the hana is also
apparent in a letter from Ibâl-El, in which he writes that ‘My lord knows that I govern the hana, and
that the hana travel through (lands in) war and peace as merchants do, and that during their movements they learn what (the people of) the lands say’ (translated from Durand 1997: t. 333).
Despite the epigraphical arguments in favour of the interpretation of hana as tent-dweller,
archaeological evidence for the use of tents during the early second millennium bc is currently
lacking. It has been argued that the tent was a relatively late phenomenon and that clear evidence
for its use only appeared during the first millennium bc. Based on circumstantial evidence, it has
for example been suggested that large woven tents such as those traditionally used by Bedouin
only appeared during the first millennium bc (Rosen 2008: 125; but see Hole 2009: 264). Although
such tents facilitate higher mobility, they are not essential for a mobile way of life. A diversity of
temporary structures can be observed among (pre-)modern Near Eastern nomadic pastoralists,
including structures made of reed, and tents that were roofed with multiple pieces of woven cloth
as well as felt and animal skins (Cribb 1991: 105–8). Many of these structures would not significantly impede mobility. Horne described temporary camps in Iran that often included permanent
features such as walls, whereby only valuable and transportable materials were brought in by the
occupants when they came to the site (1993: 45–7). Mobile groups travelling along fixed routes or
within a circumscribed region may have maintained several of such stations at preferred camping
locations. The absence of clear evidence for tents during the early second millennium bc, then,
could suggest that the translation of the term hana as tent-dweller may not be entirely appropriate,
but this absence cannot be seized upon to disregard the strong focus on mobility that is evident
in the hana way of life.83
In order to understand the position of the hana in the Mari state, it is necessary to look in detail at the organization of the kingdom’s power structure. In fact, under the king there were two
separate power structures. The first of these was concerned with the rule of the ‘districts’ (halsum)
82 Durand’s translation of the term hana as Bedouin is extremely unfortunate, as the term Bedouin is associated with certain cultural and social features that may not have been part of Mari society, and that suggest direct continuity between
hana and Bedouin. Fleming wrote on this issue: ‘Dominique Charpin brings to my attention the fact that French does
not offer the possibility of a compound word that follows the proposed etymology, as in the English “tent-dweller.”
This means that “bédouin” must serve to designate the same general category, without any forced anachronism by
association with peoples from the modern Middle East’ (2004: 252, n. 67).
83 In fact, one text from Mari mentions that the hana are in the steppe with their livestock and tents (Durand 2004a:
385).
115
of the ‘Banks of the Euphrates’ (Ah Purattim). These districts were headed by governors (šāpitum)
that were identified by town and ruled from a local palace. Individual towns in the Ah Purattim
and beyond were led by ‘leaders’ (sugāgum) (Fleming 2004: 25).
The second power structure focused on the hana. Sim’alite hana were organized in a number
of ‘divisions’ (gayum). The term gayum was generally only applied to hana Sim’alites, although some
Sim’alites living in towns were also identified by gayum (Fleming 2004: 43–4). It seems that the term
gayum was applied to several types and levels of divisions, so that it is difficult to define a more
precise meaning of the term. Furthermore, no political functions seem to have been associated
with the gayum, as these functions were carried out by officials called merhûm and sugāgum (Fleming
2004: 43). The sugāgum were lower-level officials or ‘leaders’ who were not identified by towns, but
by tribal affiliation. Above the hana sugāgum stood two ‘chiefs of pasture’ (merhûm). These merhûm
reported directly to the king.
The position of sugāgum existed in the power structure of the Ah Purattim, as well as the structure of the hana. It seems that a further distinction existed between sugāgum based on their tribal
affiliation. Thus, Sim’alite sugāgum were primarily identified by their membership of a gayum, although they were not the leaders of these divisions. Yaminite sugāgum on the other hand were often
identified by town. Rather than assuming that this dichotomy reflected a difference in the degree
of sedentarization of Sim’alites and Yaminites, it is better to assume that there were organizational
differences between the two confederacies. Furthermore, it is possible that sugāgum ruled towns,
even when they were not identified by these towns (Fleming 2004: 51).
The relation between hana, sugāgum, and merhûm is clarified by a letter about a deserter from
Zimri-Lim’s army, quoted by Fleming (2004: 50, FM III 136): ‘If [(my lord?) does not arrest(?)]
the one who takes leave (of his troop), the life of the Hana will take leave (as well). He will break
the fortune(?) of the Hana. Now then, Lawasum, the Yumahammû, has taken leave. His sugāgum is
Dadi-[Lim], and his property consists of 200 sheep and five donkeys. My lord should confiscate
(this).’ The ordering of the political structures of which the deserter was a part reveals that he was
first of all a Sim’alite hana, thereby excluding Sim’alite townspeople, that he was a Yumahammû,
which is known from other texts to be a gayum, that his sugāgum was Dadi-Lim, and that he owned
sheep, rather than land or a house. Thus, there was a clear distinction in how the different groups
that constituted the Mari state were governed, that is, sedentary townspeople on the one hand, and
(semi)nomadic pastoralists or hana on the other hand (Fleming 2004: 25).84
Despite the strong evidence in the Mari archives for a highly mobile population component
that is involved in specialized sheep/goat pastoralism, the composition of this group is much less
clear. It is clear that the term hana designates a specific group within larger tribal divisions, and
that the term is primarily defined through its association with pastoralist activities. The inter-tribal
relation between mobile hana and their sedentary counterparts is however much more difficult to
define. Due to the large variety in pastoralist modes of existence, and in the associated degree of
mobility, it is difficult to determine exactly whether the term hana designated entire households
that were fully mobile, or fully or semi-mobile parts of households.
Nevertheless, it is suggested here that the term hana designated entire households that were primarily engaged in pastoralist subsistence activities, and that pursued a (semi-)nomadic way of life.
Specialization in pastoralism thus occurred at the level of the household, rather than on the basis
of traits such as gender or age within households. The strongest evidence for this interpretation is
the marked duality in the power structure of the Mari state, in which there are different officials
for the sedentary and the hana population components. If members of a particular household split
off to become nomadic for part of the year, they would be governed under different systems, depending on the time of the year. The duality is also and perhaps especially evident in the military
organization of Mari. It has long been recognized that hana played an important role in the army,
and that they were often separately counted and listed (Matthews 1978: 95; Charpin 2004: 87–9).
84 The term sugāgum that can be roughly translated as leader seems to have designated a number of different positions
within the power structure of the Mari kingdom. These positions varied not only according to their duties, but also to
their tribal affiliations (Fleming 2004: 63–8). According to Fleming (2004: 53), sugāgum of towns paid levies in silver,
whereas hana sugāgum paid with sheep.
116
This would have been much more difficult if the distinction between hana and townsmen was
blurred or based on a seasonal cycle. Finally, a letter from Ibâl-El to Zimri-Lim describes the nature of the exchanges taking place between the hana and more sedentary agriculturalist. The letter
describes the symbiotic relationship between several kings in the Upper Khabur area and the hana
as follows: ‘The hana have always been your pastoralists, whereas you [the kings] are their farmers’
(translated from Guichard 2002: 160). Apparently, the kings had not given the hana the grain deliveries to which they were entitled, because Ibâl-El settled the matter by giving grain and oil rations
to them (Guichard 2002: 160). This letter again emphasizes the distinct nature of the different
population components, and also suggests that transactions taking place between them may have
occurred on the basis of generalized reciprocity (Guichard 2002: 164; Durand 2004b: 191).
A shift in the nature of the evidence for pastoralism has been observed to occur from the mid–
late third millennium to the early second millennium (see Section 6.4). For the earlier period, the
faunal material is dominant, whereas for the second millennium, the number of sites with faunal
assemblages is rather limited, and textual evidence is much more abundant. It is important, therefore, to consider the possibility that the reconstructed late third–early second millennium bc shift
in the organization of specialized pastoralism does not represent a historical development, but is
recognized as result of changes in the nature of the available evidence. Although the early second millennium faunal material is relatively scarce compared to the earlier periods, no significant
changes seem to have occurred in the composition of the range of domestic animals that were
kept at sites. It is clear, however, that epigraphical as well as settlement evidence indicate changes
in the organization of pastoral production during the late third–early second millennium bc. The
changes documented by the respective types of evidence cannot presently be definitively linked to
each other, but the Mari evidence clearly documents the result of this transition.
6.6 Conclusions
The present chapter has highlighted the evidence for pastoralism in the Near East in general, and
in northern Mesopotamia in particular. Animal husbandry has been practised since the late PPNB,
but the organization of this subsistence strategy has proved much more difficult to reconstruct.
It seems that at various points in time, and as a response to different environmental and sociocultural conditions, pastoralism was spatially separated from areas where agriculture was practised
in order to meet higher subsistence requirements.
During the mid-third millennium bc, this process seems to have taken place in the Upper
Khabur area, whereby the Middle Khabur and Jebel Abd al-Aziz regions functioned as overflow
regions where large-scale pastoralism could be practised without encroaching upon agricultural
production. This specialized pastoral subsistence strategy was practised out of relatively large
settlements around the Jebel Abd al-Aziz, that took advantage of local favourable environmental conditions and where agriculture was practised only in a subsidiary fashion. At the same time,
however, small-scale mixed animal husbandry was carried out at urban sites, possibly by individual
households and aimed at household subsistence production. For the late third millennium, the faunal assemblages from sites and textual evidence suggest that this dual economy continued to exist,
although evidence on the exact nature and localization of specialized pastoralism is lacking.
During the early second millennium, the Mari evidence suggests the existence of a highly mobile population component involved in pastoral production. Although second millennium bone
assemblages from northern Mesopotamia are relatively scarce, it seems that sheep/goat remained
the dominant domesticated animal. Sheep/goat flocks were moved over relatively large distances,
and the texts indicate that pastoralists were recognized as a distinct group within tribal confederacies. The absence of settlement evidence comparable to the mid-third millennium Kranzhügel
suggests that if the hana maintained settlements or camps in the nawûm areas, these settlements
were much more ephemeral than before, again suggesting a higher mobility than existed in earlier periods. Alongside this mobile pastoralism, village-based herding continued to be practised,
as possibly evidenced by the occurrence of the terms hasiratum and rubsatim, referring to ‘camps’
and ‘pens’. Thus, it is possible to reconstruct for the early second millennium the dual pastoral
117
economy that existed previously, but with a change in the organization of the fully specialized
subsistence strategy. The archaeological settlement evidence from the Wadi Ağiğ region seems to
support this reconstruction (see Section 5.4.1). There, third millennium bc material was conspicuously absent from the region, whereas second millennium bc material was relatively scarce. The
spatial patterning of this material resembled that of the Medieval Islamic period, when nomadic
pastoralists certainly visited the region. Thus, in this area that was certainly within the influence
sphere of Mari, evidence for mobile occupation in the early second millennium appeared after an
absence of some thousand years.
The present chapter thus argues for the existence of two separate pastoral economies, distinguished by the degree of specialization. It is unlikely that there was ever a fully specialized
agriculturalist or pastoralist way of life. Rather, each strategy was implemented as dictated by
environmental and social conditions. In each subsistence system, agriculture as well as pastoralism was practised, but the degree of specialization differed, as well as the degree to which the
subsistence strategy could meet the demands of the social group. At least from the mid-third millennium onwards, a dual pastoral economy consisting of a highly specialized component and a
component representing pastoralism practised in a subsidiary way can be reconstructed. This dual
economy fluctuated not only in terms of the relation between the two components, but also in
the organization of each component. Whereas the pastoral production carried out by agriculturalists was geared toward local consumption, the specialized pastoral production was geared toward
the production of exchangeable products, notably wool, though the potential importance of dairy
products should not be underestimated. Whereas this distinction can be identified for the mid
to late third millennium, in the absence of economic administrative archives of the Mari state, it
remains to be determined whether wool production was also the economic rationale for the hana
pastoralists.
118
7 The development of social networks
7.1 Introduction
This chapter will focus on exchange networks as a mechanism against environmental stress. It has
already been argued that the geographical configuration of such networks is not random. More
specifically, exchange networks will preferentially develop between groups and/or regions relying on different resources, or experiencing different environmental conditions (Section 3.5.2). In
other words, exchange networks can only effectively buffer against environmental stress if their
geographical expansion exceeds the area where stress is experienced, or if they forge ties between
groups practising different subsistence strategies. These exchange networks can potentially be reconstructed through their social component, that is, the identity or components thereof likely to
be shared by the network participants (Section 3.6). Historical records provide details on the constellation of northern Mesopotamian society in the form of the names of various social groups
that lived there during the third and early second millennium bc. By analyzing the social nature of
these groups, it can be determined whether or not they represented identities functioning in the
way that was hypothesized in Section 3.5.3.
There is one Mesopotamian identity that has the greatest potential for functioning as the social
expression of an exchange network, namely the Amorite identity. The historical records make clear
that the group called the Amorites existed in numerous regions of the Near East and was involved
in various activities and subsistence activities, thereby fulfilling the two main criteria for functioning as a social expression of an exchange network. The present chapter will start with an evaluation of the main historical sources on the Amorites (Section 7.2), followed by a discussion of how
the Amorite identity should be understood, and whether or not this identity actually functioned
as the social expression of an exchange network (Section 7.3). This chapter does not aim to give a
complete overview of the evidence for the existence of Amorites in greater Mesopotamia. Rather,
the focus will be on those texts that highlight the nature of Amorite identity, their perception of
themselves and how they were perceived by others.85
7.2 The Amorite identity in the textual record
The term Amorite is the English translation of several loosely connected terms in ancient Near
Eastern languages. It is thought to derive from the Sumerian martu through the Akkadian amurru
and the Hebrew `ěmōrī. These terms had several interconnected meanings, complicating interpretations of texts in which the terms were used. At different times and in different contexts, the term
martu/amurru could be used to designate the compass direction west, a specific area, and/or a people. How each of these uses evolved, and which meaning led to the others, is not entirely clear, as
the different meanings are related and probably evolved out of each other (Buccellati 1966: 361–2;
Whiting 1995: 1231–2). Any differences in meaning that may have existed between martu and
amurru, and between different periods, are even further conflated by the translation of these terms
into a single English term Amorite. The issue is complicated further by the fact that Amorite does
not survive as a written language. What is known of the Amorite language primarily stems from
names that are thought to be Amorite, and from rare words and expressions in Akkadian texts
(Whiting 1995: 1233). Despite these uncertainties, the term Amorite will be retained here, partly
because no adequate alternative is currently available, and partly because the present study deals
85 See Buccellati 1966, Gelb 1980, and Streck 2000, 2004 for statistical analyses of the occurrences of Amorite names
and words in historical texts.
119
date bc
period/reign
description
1792–1750
Hammurabi
Amorite genealogy, claimed Amorite affiliation
1810–1761
Mari
Mari as Amorite state under Yahdun-Lim, Shamshi-Adad/Yasmah-Addu, and
Zimri-Lim
1830–1776
Shamshi-Adad
Amorite genealogy
1834–1823
Warad-Sîn
Warad-Sîn’s father Kudur-mabuk was called ‘father of the Amorite land’
1941–1933
Zabāia
ruler of Larsa claimed Amorite affiliation
2019–1987
Išbi-Erra
described as ‘a man from Mari, with a dog’s intelligence’
2026–2003
Ibbi-Sin
correspondence between Ibbi-Sin and Išbi-Erra concerning grain transports and
Martu activities
2032–2031
Šu-Sin 4–5
year name mentioning construction of the murīq-tidnum
2032
Šu-Sin 4 (terminus post quem)
correspondence between Šu-Sin and Šarrum-bani concerning the murīq-tidnum
2056–2055
Šulgi 37–38
year names mentioning construction of the ‘wall of the land’
2056
Šulgi 37 (terminus post quem)
correspondence between Šulgi and Puzur-Šulgi concerning the ‘wall of the land’
2056
correspondence between Šulgi and Arad-ĝu concerning the ‘wall of the land’
2064
Amorite names in administrative records
c. 2100
Gutian/Ur III?
Curse of Agade: Amorites are described as ‘people ignorant of agriculture’
c. 2100
Gudea
transport of building materials from the mountain range of the Martu
c. 2160
Šar-kali-šarri
year name mentioning victory over Martu at Bašar
Naram-Sin
victory over Amar-Girid of Uruk at Bašar; capture of Martu leaders
c. 2350
Tell Beydar
Amorite names (?)
c. 2350
Ebla palace G
various interactions between Ebla, Martu, and Tidnum (?)
c. 2500
Lagash (ED III)
Amorite toponyms
c. 2500
Shuruppak (ED IIIa)
Amorite personal names
Table 7.1: Selected textual references to Amorites in chronological order. See main text for references.
primarily with texts from the third and early second millennium bc, thereby circumventing some
ambiguities that were introduced by the slightly later use of amurru to refer to a northern Levantine
state, rather than to a group of people (Klengel 1992: 160; Whiting 1995: 1236).
Due to the uneven spatial and chronological distribution of the available evidence, the present
review includes historical records from greater Mesopotamia (Table 7.1). For the purpose of the
present overview, the mid–late third millennium and early second millennium will be divided in
three periods, for which clear differences in the nature and availability of the evidence can be
demonstrated. The first period includes all evidence predating the Ur III period. This evidence
is meagre and sometimes rather ambiguous. The middle period coincides with the Ur III period,
when there is much more evidence on Amorite presence and identity, although most evidence
comes from southern Mesopotamia. The final period covers the early second millennium and
roughly corresponds to the Isin-Larsa/Old Babylonian periods. The degree to which differences
in the available evidence influence the interpretation of the texts will be addressed below (Section
7.3.1).
7.2.1 Amorites before the Ur III period
Steinkeller has suggested that the term Tidnum, which is used nearly synonymously for Amorite
during the Ur III period (see Section 7.2.2), possibly already occurs in archaic ED I texts from
Ur (1992: 265), but this suggestion cannot be further substantiated at the moment (Marchesi
2006: 22–3). The first reliable references to people of supposedly Amorite affiliation date to the
ED IIIa period and come from texts found in Shuruppak. There are a few names of supposedly
Amorite origin, despite the fact that the determinant Martu is lacking (Bauer et al. 1998: 265, n.
268). However, this qualification does appear at this site in combination with personal names,
for example in a list of barley rations: ‘900 litres of barley for Esur’agga, the Amorite’ (Marchesi
2006: n. 95). Another text reads ‘45 men receiving 1 loaf of bread (each), (and) 28 women receiv-
120
The development of social networks
ing one loaf of bread (each) – (they are) Amorites’ (Marchesi 2006: n. 95). In addition, a ‘land of
the Amorites’ is mentioned in a literary composition (Marchesi 2006: n. 95). In ED III texts from
Lagash, several toponyms occur in association with Martu. An inscription from Ur-Nanshe mentions the digging of a ‘canal of the Amorites’ (Marchesi 2006: 23–4; Frayne 2008: E1.9.1.31).86
References to a ‘bridge of the Amorites’ occur several times in these texts (Marchesi 2006: 24).
A pre-Sargonic literary text from Mari, possibly based on an earlier but otherwise unknown text
from Lagash, mentions an ‘Amorite shepherd’ (Marchesi 2006: n. 99 quoting Bonechi and Durand
1992).
The Ebla palace G archives contained several references to Amorites and an Amorite region
(Archi 1985). An inhabitant of Tuttul (Tell Bi’a) was called a resident of Martu (Archi 1985: text
30). Inhabitants of Emar were in some way involved in a victory over Amorites: ‘(fabrics) for
Aršum of Emar, for a votive offering (when/because?) Martu was destroyed, and for one of
(Aršum’s) sheep for the victory. (Fabrics) for Baluzú of Emar, when Martu was destroyed’ (Archi
1985: text 9). There were Amorite rulers and councils of elders who each received gifts of cloth
(Archi 1985: text 14). Several texts recorded the delivery of sheep by people from Martu. Finally,
there was a list including personnel from Martu. Note that individuals designated as coming from
Martu(m) never had Amorite names. A possibly older variant of the term Tidnum has also been
attested in the Ebla archives (Marchesi 2006: 11).87
There is discussion about the occurrence of Amorite personal names in the roughly contemporaneous texts from Tell Beydar. According to Talon, there were no Amorite names or Amorite
elements in the Beydar text corpus (1996: 75). However, Streck (2002: 191) suggested that several
names may have been Amorite.
Amorites were mentioned by Naram-Sin and his successor Šar-kali-šarri. Naram-Sin’s account
of the Great Revolt described how he crossed the Euphrates and fought the decisive battle against
Amar-Girid of Uruk at Bašar, the mountain of Martu. After the battle, Naram-Sin captured AmarGirid and, among many others, two persons named Belili and Kindagal, who were described as
leaders of the Martu people (Sommerfeld 2000: 423–6). Naram-Sin’s successor Šar-kali-šarri was
the second Akkadian king to mention the Martu. One of Šar-kali-šarri’s year names recorded a
victory over the Martu at Bašar: ‘The year in which Šar-kali-šarri defeated the MAR.TU in Bašar’
(Sommerfeld 2000: 435).
The Curse of Agade, probably to be dated within 100–200 years after the Akkadian period
(Cooper 1983: 11–12), described how the Amorites submitted to Naram-Sin and offered to the
goddess Inanna: ‘The highland Martu, people ignorant of agriculture, brought spirited cattle and
kids for her’ (Black et al. 1998–2006: t.2.1.5).
Gudea statue B described, among other things, the building of the Ningirsu temple, and paid
close attention to where building materials came from: ‘From Umanum, in the mountain range of
Menu’a, and from Pusala, in the mountain range of the Amorites, he (=Gude’a) brought down
big stone slabs, … From Tidanum, in the mountain range of the Amorites, he brought alabaster
blocks’ (Marchesi 2006: 16). According to Marchesi, Pusala and Tidanum were two different places
in the mountain range of the Amorites, rather than being two synonyms for the mountain range
of the Amorites, as had been suggested earlier.88
86 Frayne remarks that it is not entirely clear whether Martu here indicates the Amorite identity or the compass direction
west. He suggests that the Amorite interpretation is plausible, based on the mentioning of LUM-ma as Eanatum of
Lagash’ Tidnum name. However, Marchesi recently argued that the link between LUM-ma, Tidnum, and Eanatum is
based on a false reading of the text, and suggested that LUM-ma should be understood as an as yet unknown ruler of
Lagash (2006: 124-5).
87 Note that mar-TUMki has recently been identified as a cult place near Ebla, possibly necessitating a revision of some
of the conclusions drawn by Archi (Sallaberger 2007: 445, n. 138).
88 Marchesi’s interpretation also contrasts with earlier readings in its reading of Pusala, which had previously been read
as Bašar, for example by Edzard ‘He brought down big stone slabs from Umānum, the mountain range of Menua,
and from Basar, the mountain range of the Martu, ... He brought alabaster blocks from Tidanum, the mountain range
of the Martu’ (1997: E3/1.1.7.StB, vi 3-14). It also contrasts with Michalowski (1976: 111) who argued that Bašar
and Tidnum were two distinct places that were each called ‘Mardu highland’. See Sallaberger (2007: n. 79) for further
discussion of the problematic reading of Gudea Statue B.
121
7.2.2 Amorites in the Ur III period
There are numerous historical records from the Ur III period (2110-2003 bc) dealing with or
mentioning Amorites. These records generally fall into one of two classes: administrative records
mentioning Amorite names or literary texts including year names, royal correspondence and other
public literature describing activities of Amorite groups, usually without mentioning individual
names. As will be seen below, these two classes of historical records present the Amorite identity
in fundamentally different ways.
The occurrence of Amorite names in Ur III administrative records has been studied by
Buccellati (1966). His study of Amorites during the Ur III and early Isin-Larsa periods provided
a picture of Ur III society in which Amorites played an active role and were involved in various
economic activities. His main sources of information were administrative texts from archives
found in Puzrish-Dagan, Isin, and Lagash. He was able to identify over 300 names with the determinant mar-tu, or with West-Semitic elements that are thought to be Amorite. He then studied the
economic and social contexts in which these names occurred, and the spatial and chronological
distribution of these names.
The greater share of the Amorite names occurred in the Puzrish-Dagan texts, which dealt
with the circulation of animals by the central administration. Amorites were recorded as delivering and receiving animals for a variety of purposes. As contributors, Amorites delivered relatively
large numbers of fat-tail sheep. Although Buccellati found no seasonal pattern in the presence of
Amorites at Puzrish-Dagan (1966: 346), it has been shown that there were seasonal fluctuations
in the number of animals that were delivered (see Section 6.4). The texts from Isin described
transactions of leather products. The majority of Amorites apparently received these products,
and, with a few exceptions, were involved neither in the production of these products, nor in the
contribution of the raw materials. Finally, the Lagash administration primarily documented the
distribution of food rations to various officials, several of whom were Amorites. The Amorite
officials primarily held lower positions within the bureaucracy. Amorites also held positions outside the state administration, including such diverse functions as solder, priest, brewer, or farmer
(Buccellati 1966: 340-2).
Buccellati cursorily referred to the textual evidence from other sites (Buccellati 1966: 316).
Thus, at Eshnunna, there were numerous references to Amorites in texts from both the Ur III and
Isin-Larsa periods. At Nippur, Amorites were virtually absent from the textual record. The texts
from Umma suggested to Buccellati that relatively few Amorites lived there. At Larsa, evidence
for an Amorite presence came from the first two names on the Larsa king list, which occurred as
Amorite names elsewhere. At Ur, attestations of Amorites were also few and far between, and they
primarily occurred in administrative texts recording the distribution of goods.
The onomastic evidence suggests that Amorites were well-integrated into the Ur III society,
and that they functioned in the state bureaucracy, as well as in other parts of society. The references to Amorites in literary sources are of a different nature. Literary sources, mainly consisting
of year-names and the letters from the Royal Correspondence of Ur, usually provide a rather onesided picture of the Amorites, using expressions and descriptions that clearly aimed to portray
Amorites as enemies of the state. The Amorites were first mentioned in year names from Šulgi’s
reign. The year names of Šulgi 37 and Šulgi 38, mentioned the construction of the ‘wall of the
land’ (bad3-ma-da) or the ‘wall facing the highlands’ (bad-igi-hursaga).
This ‘wall of the land’ also played a significant role in the so-called Royal Correspondence of
Ur, a body of letters between several Ur III kings and high officials within the state (Michalowski
1976). It should be noted, however, that several researchers have questioned the historicity of the
so-called Royal Correspondence of Ur, of which the letters from or to the Ur III rulers quoted
here are a part. These letters are known only from Old Babylonian copies, in other words, they
were made at a time when fundamental changes had occurred in the socio-political constellation
of southern Mesopotamia as compared to the Ur III period. The problem is that it is difficult, if
not impossible, to distinguish historical details from parts that were added or invented when the
texts were composed or copied. Michalowski, however, has shown that personal names in the royal
122
correspondence of Ur also occurred in genuinely Ur III texts, suggesting that the texts at least incorporated some historical details (Michalowski 1976: 65, 67). Therefore, it is perhaps best to treat
these texts as secondary historical sources (Streck 2000: 36; Charpin et al. 2004: 59, n. 145).
A letter from the high official Aradĝu to Šulgi referred to the construction of the wall: ‘As
to the fortification which my lord sent me back to, the work on it has been put into effect. The
approach of the enemy is kept at a distance from the Land. My lord continues to maintain his
sublime reputation in the south and the uplands, from the rising to the setting sun, as far as
the borders of the entire Land. {The rebellious (?) Martu have turned back ……}’ (Black et al.
1998–2006: t.3.1.06). The correspondence between Šulgi and another official, Puzur-Šulgi, was also
concerned with the construction and restoration works of the ‘wall facing the highlands’. After boasting that he had built the wall and brought peace and prosperity to the land, Šulgi wrote: ‘When the
master-builder (?) has taken up the work concerned, he is to re-establish securely any place where
the fortification has fallen into ruins. Let him reinforce and also rebuild it. The neglected work
load is to be completed within one month; I shall be questioning him about this work.’ (Black et al.
1998–2006: t.3.1.08, version A, segment B). The urgency of the task was again stressed at the end
of the letter: ‘The orders are rigorous: you should not neglect your work load. They are to proceed
with the building work by night and in the heat of noon. You will not be sleeping during the night
or in the heat of noon! You should know this! It is urgent!’ (Black et al. 1998–2006: t.3.1.08, version A, segment B). The same letter also mentioned that the ‘…… the Tidnum have returned (?)
from their mountain land’ (Black et al. 1998–2006). In another letter, Puzur-Šulgi wrote to Šulgi: ‘I
am also well-informed about the oracular signs concerning the enemy: the enemy has replenished
his strength for battle. However, my strength is limited. I cannot strengthen the fortress further or
{guard it}’ (Black et al. 1998–2006: t.3.1.07). He furthermore suggested that Šulgi send him 7200
workers for the construction of the wall (Black et al. 1998–2006: t.3.1.07).
The mentioning of the Tidnum in close association with the ‘wall of the land’ is important,
because it establishes a link between the Tidnum and the Amorites as groups that are apparently
in some way related. In different copies of texts belonging to the Royal Correspondence of Ur,
Tidnum and Martu are used synonymously for both a group of people, and a geographical place
name, suggesting at least a considerable overlap in meaning. This suggestion is further enhanced
by several lexical lists that equate Tidnum with Martu (Marchesi 2006: n. 20, 23). At the same time,
a text from Šu-sin describing a battle indicates that Tidnum could also be used in a more restricted way to describe a particular group of Amorites: ‘Hostile Amorites – Tidnum and Yamatiyum
– came forth (to do battle): their kings confronted him (=Šusîn) in battle, (but) by the strength of
Enlil and Ninlil he was victorious in battle’ (Frayne 1993b: E3/2.1.4.1; Marchesi 2006: 12).
During the reign of Šu-sin (2035-2027 bc), there was again work on a wall as a defence against
Amorite attacks. Šu-sin’s fourth and fifth years were named after the construction of this wall.
The official Šarrum-bani, who had been ordered to build or reconstruct the wall, wrote to Šu-sin:
‘You sent me a message ordering me to work on the construction of the great fortification MurīqTidnim. {You presented yourself before me} […], announcing: “The Martu have invaded the
land.” {You instructed me} […] to build the fortification, so as to cut off their route; also, that no
breaches of the Tigris or the Euphrates should cover the fields with water’ (Black et al. 1998–2006:
t.3.1.15). This letter also indicated that Šarrum-bani had been building and fighting at the same
time. Šu-sin replied that Šarrum-bani had not been instructed to fight, but only to reconstruct
the wall that Šulgi had built (Black et al. 1998–2006: t.3.1.16), indicating that the ‘wall facing the
highlands’ (bad3 igi-hursaga) and the ‘wall which repels the Tidnum’ (bad3 mu-ri-iq-tidnim) were
probably the same, or at least located in the same region.
In the absence of archaeological evidence for the ‘wall of the land’, its location can only be
reconstructed from the textual evidence. Michalowski has shown that the ‘wall of the land’ was
usually mentioned in association with place names that can be located in or near the Diyala region.
More specifically, the wall is described as running between the Abgal waterway and Zimudar, and
123
having a length of 26 danna (Black et al. 1998–2006: t.3.1.15).89 The Abgal waterway has been
equated with the western branch of the Euphrates at Sippar, whereas the location of Zimudar is
uncertain but probably lies somewhere in the Diyala region (Michalowski 1976: 114). This location
in the Diyala region and the fact that the ‘wall which repels the Tidnum’ was also called ‘wall facing
the highlands’ suggest that the wall was built toward the Jebel Hamrin (Michalowski 1976: 115).
The nature of the wall remains unclear. It is unlikely that it was a large continuous wall, and a chain
of small fortifications or watchtowers has been proposed as a more likely alternative (Buccellati
1997: 109).90
A letter by Išbi-Erra, the later ruler of Isin (2019-1987 bc), to the last Ur III king Ibbi-Sin
(2026-2003 bc) indicated that Išbi-Erra was instructed to purchase grain, but was forced to abandon this expedition because of the presence of Martu: ‘I heard news that the hostile Martu have
entered inside your territories. {I entered with 72,000 gur of grain} […] – the entire amount of
grain – inside Isin. Now I have let the Martu, all of them, penetrate inside the Land, and one by
one I have seized all the fortifications therein. Because of the Martu, {I am unable to hand over}
[…] this grain for threshing. They are stronger than me, while I am condemned to sitting around.’
(Black et al. 1998–2006: t.3.1.17). The letter furthermore indicated that Elam, with which the Ur
III state was also in conflict, was likewise suffering from grain shortages (Black et al. 1998–2006:
t.3.1.17).91
In another letter from the Royal Correspondence of Ur, Ibbi-Sin described Išbi-Erra as ‘not
of Sumerian origin’ and as a ‘man from Mari, with the understanding of a dog’ (Black et al.
1998–2006: t.3.1.20). Cooper noted that the qualification ‘with a dog’s intelligence’ was usually
applied to Amorites and Gutians, but not to foreigners in general because Elamites were almost
never described as such, leading him to doubt the Babylonian origin of Išbi-Erra (1983: 33). The
Mari origins of Išbi-Erra were also highlighted in a hymn known from Old Babylonian copies,
but probably originating from the time of Išbi-Erra, that described how the gods Anu and Enlil
elevated Išbi-Erra to kingship (Michalowski 2005).
Amorites also appeared in several literary compositions. In the well-known literary composition The marriage of Martu, the god Martu is described as follows (Black et al. 1998–2006: t.1.7.1):
‘Now listen, their hands are destructive and their features are those of monkeys; he is one who
eats what Nanna forbids and does not show reverence. They never stop roaming about ……, they
are an abomination to the gods’ dwellings. Their ideas are confused; they cause only disturbance.
He is clothed in sack-leather ……, lives in a tent, exposed to wind and rain, and cannot properly
recite prayers. He lives in the mountains and ignores the places of gods, digs up truffles in the
foothills, does not know how to bend the knee, and eats raw flesh. He has no house during his life,
and when he dies he will not be carried to a burial-place’. Note, however, that this composition
does not describe the Amorites as such but the god Martu/Amurru, necessitating an investigation
of the link between this god and the Amorites. It has been observed that Martu/Amurru does
not occur as a theophoric element in Amorite names, but that it does occur in Akkadian names.
Beaulieu therefore suggested that the god Martu/Amurru was a Sumerian/Akkadian construct ‘on
the basis of their perception of a separate ethnic identity’ (Beaulieu 2005: 35). As such, the god
Martu/Amurru may have served as a stereotypical image of the Amorites in various aspects of
Mesopotamian society.
89 The exact length of 1 danna is not entirely clear but may range between 7.3 and 11.5 km, with 10.8 km being most
probable (Powell 1990: 467), making the wall between 190 and 300 km long, and 280 km if 1 danna≈10.8 km is
accepted.
90 Sallaberger rightly points to the low wall that has recently been found near Al-Rawda and that may eventually shed new
light on the ‘wall facing the highlands’ (2007: n. 136)
91 During the Ur III period, 1 gur=300 sila, and 1 sila≈1 litre (Powell 1990: 493), so that 72000 gur would be 21600000
litre. It should be noted however that this letter is not an administrative, but a literary text, making the accuracy of
these figures doubtful.
124
There is evidence from the Ur III period that Amorite existed as a spoken language. A literary
text mentioned an ‘Amorite interpreter’, and Šulgi claimed to be able to speak Sumerian, Elamite,
and Amorite. Letters from Old Babylonian Mari recognized Akkadian, Sumerian, Subarean, and
Amorite as separate languages (Streck 2000: 76). Nevertheless, Sumerian and later Akkadian remained the dominant written languages.
For northern Mesopotamia, there is some evidence for Amorite names in texts from Mari and
Tell Bi’a. Several of the latest Mari šakkanakkus bore Amorite names and words like gayyum (clan)
were also already present (Streck 2004: 340).
7.2.3 Amorites in the early second millennium
bc
The Ur III period and the šakkanakku period in Mari ended shortly before 2000 bc, after which
documentation on the presence of Amorites in Upper Mesopotamia ceased, only to start again
some 200 years later. This is unfortunate, since this period must have witnessed significant changes,
including the appearance of Amorite kingdoms across Mesopotamia. Thus, Charpin and Ziegler
(2003: 29) called this period ‘la grande vague’ and argued that it witnessed large population displacements or ‘invasions amorrites’. By the time this epigraphic silence had been broken, the political landscape of greater Mesopotamia had changed considerably. At the same time, the Sumerian
determinative mar-tu to designate Amorite persons disappeared almost completely. Buccellati observed that this development had already commenced during the Ur III period (1966: 358). After
this time, Amorites could only be distinguished through their names, or if they specifically stated
their Amorite affinity.92
During the Old Babylonian period, numerous city-states were now ruled by rulers with Amorite
names, or who claimed Amorite descent. Amorite dynasties became established at Babylon,
Eshnunna, Isin, Larsa, Mari, Qatna, and Yamhad. Several of these kings specifically mentioned
their Amorite affiliation. Thus, Zabāia, fourth king of Larsa (1941-1933 bc), wrote: ‘Zabāia,
Amorite chief, son of Sāmium, built the Ebabbar’ (Frayne 1990: E4.2.4.1). The title ‘Amorite chief ’
or ‘commander’ was also used by Abi-Sare of Larsa (1905-1895 bc) and Sin-gamil of Diniktum
(Michalowski 1976: 121). Kudur-mabuk, the father of Warad-Sîn of Larsa (1834-1823 bc), was
called ‘father of the Amorite land’ (Frayne 1990: E4.2.13.13, l. 14-15). Note the kinship element
present in this title. Hammurabi of Babylon called himself ‘king of all the Amorite land’ (Frayne
1990: E4.3.6.8) and ‘king of the Amorites’ (Frayne 1990: E4.3.6.2001).
Some kings apparently even claimed or constructed a shared Amorite ancestry, that is, ShamshiAdad of Ekallatum and Hammurabi of Babylon. Comparison of their genealogies, as recorded
by the Assyrian King List (AKL) and the Genealogy of the Hammurabi Dynasty (GHD), reveals
some interesting patterns (Finkelstein 1966). Several names of rulers preceding the first dynasty
of Babylon on the GHD also appear on the AKL, although in a different order. These names
are Heana/Hanū, Ditānu/Didānu, and Zummabu/Zu’abu. In the AKL, these kings were among the
seventeen kings who lived in tents. Ditānu/Didānu is known as an Amorite group, whereas Heana/
Hanū is a term that is closely associated with pastoralism is general (see Section 6.5.2). Two later
entries in the GHD are very similar to the names of Amorite groups living in Sippar. Finkelstein
furthermore suggested that several personal names in the AKL appeared in a fused and/or distorted way in the GHD, and vice versa, eventually concluding that ‘the genealogical traditions of
the Hammurapi dynasty and those of the Assyrian King List – the first two sections of which must
almost certainly be identified as the “Ahnentafel” of Šamši-Adad I – are one and the same insofar
as they represent a consciousness of tribal origins’ (1966: 99).
The structure of the emerging Amorite kingdoms is most clearly visible in the archives from
Mari, supported by several smaller archives from other northern Mesopotamian sites, including
Tell Leilan, Chagar Bazar, and Tell al-Rimah. These archives primarily document a relatively short
time-period centred on the reigns of Shamshi-Adad and Zimri-Lim. The Mari archives, being
the largest among these text corpuses and containing c. 20000 complete and fragmented tablets,
92 The appearance of Amorite dynasties across Mesopotamia and western Syria led Charpin to suggest that the Old
Babylonian period could more appropriately be named the Amorite period (2004: 31).
125
consist predominantly of administrative documents relating to the functioning of the palace, and
letters dealing with a variety of subjects, from which state affairs can be reconstructed in unprecedented detail. Precise data on the archaeological contexts are lacking for the texts found in the
early excavations by Parrot, but some patterns have emerged from the more recent excavations at
Mari. These indicate that the majority of texts pre-dating the reign of Yasmah-Addu were probably discarded or buried in the past, possibly indicating a lack of interest on Yasmah-Addu’s part
for keeping alive the remembrance of earlier rulers. The archives from the reign of Zimri-Lim, on
the other hand, contained letters from the reign of his predecessor, perhaps indicating a stronger
historical consciousness (Durand 1997: 29). Nevertheless, it should always be kept in mind that
these texts were part of the palace archives, and written by palace officials. The views expressed in
them therefore were the views of the state, and the matters discussed in them were of importance
to the state.
Before the structure of the Mari state is more fully explored, it may be useful to clarify the use
of the terms tribe and tribal, and whether these are applicable to the society that is documented in
the Mari archives. This study will follow the proposals made by Fleming (2004: 26), building on earlier work by Salzman (1978a) and Khoury and Kostiner (1990a). In this approach, the term tribal is
used to describe a type of social organization usually characterized by real or constructed kinship
ties, and by shared cultural traits such as language and customs (Khoury and Kostiner 1990b: 5). A
tribal confederacy represents a coalition of tribes that is politically united, although this does not
necessarily involve a centralized government system (Fleming 2004: 27). Within this approach, the
term tribe is devoid of the evolutionary connotations evident in the works of Sahlins and Service.
Thus, a tribal organization is not a priori more advanced or primitive than other archetypical forms
of political and social organizations such as the state or the chiefdom (Parkinson 2002: 9).
It appears that Mari society was organized along lines of tribal affiliation. Rowton already realized the importance of tribal affiliations in Mari and used the terms ‘dimorphic chiefdom’ or ‘dimorphic state’ to describe a polity uniting tribal (semi-)nomadic groups and non-tribal sedentary
people living in towns (Rowton 1973a).93 In Rowton’s model of the Mari state, the state was essentially non-tribal in nature, headed by a non-tribal ruler who was assisted by primarily non-tribal
officials. The state continuously tried to control the tribal elements that were present within its
jurisdiction. These tribal groups, however, actively resisted and, because they were (semi-)nomadic,
could and often tried to evade state control by moving into the steppe beyond the state boundaries.
However, due to the cyclical nature of resources, (semi-)nomadic pastoralists always had to return
to the more fertile areas controlled by the state, and this cycle of intrusion and retreat was central
to Rowton’s notion of enclosed nomadism (Rowton 1973b, 1974). Rowton built his model of the
structure of the Mari state on observations of pre-modern and historical tribe-state interaction in
the wider Near East. Central to his model of the dimorphic state was the idea that town and tribe
were both governed by a single ruler or state administration. Thus, according to Rowton (1973a:
202), ‘The hallmark of dimorphic structure is an autonomous chiefdom centered on a town in
tribal territory. From this base a local dynasty exerts a varying blend of rule and influence over the
nomadic and sedentary tribes in the countryside. The population of the chiefdom includes both
a nontribal and a tribal element.’ Within this structure, the power of tribal leaders fluctuated, but
these leaders were eventually absorbed into or replaced by the essentially non-tribal state organization (Rowton 1974: 22).
However, recent research has indicated that the tribes or tribal confederacies were even more
strongly interwoven with the Mari state than assumed by Rowton or Matthews (1978). The Mari
state was not above or on a par with the tribes, but was itself entirely tribal. Thus, rather than assuming that the Mari rulers dealt with the tribal confederacies through intermediate (non-)tribal
officials such as the merhûm, the assumption now is that the rulers themselves were members of
one of these confederacies, the Sim’alites (Charpin and Durand 1986: 150). Within this view, the
absence of indications for a tribal identity should not be understood as an absence of tribal affiliations, but as an indication that these affiliations were so obvious that they were not always
93 Rowton has written extensively on pastoralism, nomadism, and tribe-state interaction in recent times, as well as during
the Old Babylonian period (see Rowton 1980: n. 2, for an extensive list of Rowton’s articles on these subjects).
126
mentioned. In other words, in many cases where all indication of tribal identity is absent, the text
probably related to the Sim’alite confederacy, because that confederacy dominated Mari politics
(see Section 6.5.2).94
Several Amorite tribal confederacies can be recognized, of which the Sim’alites ‘of the left
hand’ and Yaminites ‘of the right hand’ were the most important (Streck 2002: 175).95 These tribal
confederacies were ruled according to different power structures. This difference is most clearly
visible in the fact that the Sim’alites were governed by a single ruler, that is, Yahdun-Lim and later
Zimri-Lim, whereas the Yaminites had five rulers. Under their single ruler, the Sim’alites were organized in gayum, of which at least twelve are known by name (Durand 2004b: 179-80). The term
gayum is often translated as ‘clan’ or ‘tribe’, whereas Fleming suggested the more neutral ‘division’
(2004: 43). No clear official position can be assigned to the gayum, and they may not have had
any political functions (Fleming 2004: 58). Apparently a very strong social relation between hana
and the gayum existed, in the sense that virtually only hana were identified as members of a gayum.
Nevertheless, sedentary individuals could identify themselves as members of a gayum as well, indicating that a gayum was not a strictly pastoralist social unit. (Fleming 2004: 54-5).
Some evidence suggests that there was an overarching division of the gayum into hana of Yabasu
and hana of Ašarugayum, to which most of the known gayum could be assigned. The Yabasu may
have been located in the western part of the Jezirah, whereas the Ašarugayum were found around
the Jebel Sinjar and along the Euphrates south of Mari (Durand 2004b: 180-3). This overarching
division seems to have primarily functioned in military contexts, and Sim’alites were otherwise always identified by their gayum (Fleming 2004: 57).
The Yaminites were divided into li’mum, five of which are known by name: the Amnanû, Rabbû,
Uprapû, Yahrurû, and Yarihû (Durand 2004b: 158). In fact, Yaminites were generally identified as
members of one of these li’mum, rather than as Yaminites. Each Yaminite li’mum was governed
by a šarrum or king (Durand 2004b: 158). Despite these organizational differences between the
Sim’alite and the Yaminite tribal confederacies, the gayum and the li’mum were very similar in terms
of membership, as they both represented the first tribal level by which individual Sim’alites and
Yaminites were identified. The Yaminite li’mum similarly consisted of mobile pastoralists and sedentary people (Fleming 2004: 61-3). Among the Yaminites, the term hibrum seems to have denoted
a mobile pastoralist population component in much the same way as hana and gayum did among the
Sim’alites (Matthews 1978: 65; Durand 2004a: 390-1; Fleming 2004: 97).96
In his discussion on enclosed nomadism, Rowton was similarly concerned with the tribal
organization as attested in the Mari archives. He drew attention to the fact that the Mari tribal confederacies could include sedentary as well as mobile population components. He used
the term ‘integrated tribe’ to describe social groups consisting of sedentary as well as (semi-)
nomadic households (1974: 2). For example, in a letter to Zimri-Lim by Hammî-ištamar, a ruler
of the Uprapû and therefore a Yaminite, one reads: ‘My lord knows that the leaders and the tentdwellers are with me at Samânum. Because they had not met with their brothers, residents of the
city, for a long time, they spent several days without moving’ (translated from Durand 1998: text
733). The same distinction is even more strongly expressed in another letter about the division of
land. According to this text, the ‘house of Awin’, part of the Yaminite li’mum Rabbû, is divided in
five sons of Awin who live in Appan, and eight sons of Awin of the hibrum of the nawûm, presumably in the meaning of pasture area (Matthews 1978: 90). These texts not only clearly indicate that
hana could be used within a Yaminite social context, but also, and more importantly, that mobile
hana and hibrum on the one hand, and sedentary townsmen on the other were part of the same tribe
and that their affiliation was described along kinship ties.
94 Consequently, as Streck rightly observed, the tribal-non-tribal dichotomy is not an apt characterization of a dimorphic
chiefdom or state (Streck 2002: n. 145).
95 Early scientific literature on Mari nomadic pastoralism recognized the hana or Haneans as a further tribal group on
an equal organizational level as the Sim’alites and Yaminites (e.g. Kupper 1957; Anbar 1991), but it has recently been
shown that this view is erroneous (see Section 6.5.2).
96 Some social groups apparently stood outside the tribal system outlined here, i.e. the habirū (see Matthews 1978:
159-62).
127
Although there were areas with mixed tribal populations, the available evidence nevertheless
indicates that attested individual tribal identities were more strongly associated with particular
geographical regions than with others. This evidence points to the existence of settlements with a
specific tribal affiliation, as well as to the location of pasture areas of specific tribes. According to
Anbar (1991: 116), the Yaminite settlements were primarily located in the districts of Mari, Terqa,
and Saggaratum. Their pasture areas were widely dispersed across northern Mesopotamia, including Yamhad, Zalmaqum, and the Jebel Bishri. Anbar distinguished the ‘Haneans’ from other tribes,
but as has become clear, references to them should be understood as references to Sim’alites. Thus,
the observations by Anbar that the ‘Haneans’ had a special connection with Ida-Marash, and that
many Sim’alite settlements could also be found there actually reinforce each other (Anbar 1991:
116-17). The smaller tribal groups were primarily found along the periphery of Mari’s influence.
Thus, the Ya’ilānum were found east of the Tigris, whereas the Numha and Yamutbal were located
in Terqa, Saggaratum, and south of the Jebel Sinjar.
It is clear, then, that there was not only a dichotomy in Mari between a mobile pastoralist population component and a sedentary population component (see Section 6.5.2), but also a tribal dichotomy. The first dichotomy opposed hana against those people associated with the ālum, whereas
the second pitted the tribal confederacy of the Sim’alites against that of the Yaminites. Although
these tribal confederacies were characterized by very different power structures, they were social
units containing and therefore uniting smaller units from both the hana and ālum population components. An important consequence of this interpretation of Mari society is that conflicts between
the Sim’alite rulers of Mari and the Yaminite confederacy were not conflicts between the desert
and the sown, but between two tribal confederacies, each consisting of households practising pastoralist or agricultural subsistence strategies (Fleming 2004: 46).
Two such conflicts are particularly well known. The first of these is recorded in Yahdun-Lim’s
foundation inscription of the Shamash-temple. In this inscription, Yahdun-Lim described how he
defeated a Yaminite coalition consisting of Amnanû, Rabbû, and Uprapû (Matthews 1978: 134):
‘La’um, king of Samānum and the land of the Uprabum, Bahlukulim, king of Tuttul and the land
of the Amnanum, Aiālum, king of Abattum and the land of the Rabbium, these kings attacked him
(Yahdun-Lim) and to aid them (also) came the army of Sumu-Ebuh of the land of Yamhad. At the
town, Samānum, the troops of the Yaminites assembled to him (Sumu-Ebuh) as a single entity and
in a mighty battle these three kings of the Yaminites he (Yahdun-Lim) put in chains.’ According to
Matthews, this conflict represented a war between the urban ruler of Mari and the tribal Yaminites.
However, because the Sim’alites were ruled by an individual ruler, whereas the Yaminites had several rulers, this text does not indicate a conflict between tribe and state, but between two tribal
confederacies. The same situation seems to have arisen when Zimri-Lim ascended the throne of
Mari and had to quell a rebellion of several Yaminite li’mum (cf. Durand 1998: 420).
Streck (2004) analysed the chronological and spatial development in the occurrence of Amorites
names in greater Mesopotamia during the Old Babylonian period (2000-1600 bc), thereby supplementing Buccellati’s work on the Ur III period. Using a database of Amorite names appearing
in texts from numerous sites across greater Mesopotamia, he was able to document the relative
importance of Amorite names in the total corpus. Because textual data from different periods
were available at several sites, it was also possible to document changes through time. Lower
Mesopotamia, the region east of the Tigris, and the Jezirah were each characterized by different diachronic developments. In lower Mesopotamia, the percentage of Amorite names in the
total corpus decreased from an average of 15 % for early sites to 4 % for late sites. In the Diyala
and Hamrin regions, the percentage of Amorite names was 23 %. The number of sites and their
chronological spread is insufficient to point to any diachronic changes. Amorite names were particularly strongly represented in the Mari region and the Euphrates upstream to Emar. Archives
from this region contained on average 78 % Amorite names. Further west, at sites like Carchemish,
Aleppo, Mishrife, and Hazor, 75 % of the names was Amorite. At sites further east, like Chagar
Bazar, Rimah, and Barri, the importance of Amorite names was much lower, ranging between
11-36 % (Streck 2004). Both Anbar (1991: 89-90) and Streck (2004: 334) pointed to the marked
differences in the occurrence of Amorite names in tribal and non-tribal contexts in the Mari archives. Although the absence of tribal affiliation should not be taken as an indicator of non-tribal
128
ascent, the results are nevertheless interesting. Amorite names made up only c. 40 % of the total
number of names known from Mari. That percentage, however, increased to over 75-80 % when
only names of tribal members were taken into account.
Streck was also able to study the occurrence of Amorite names within single families and over
several generations in the archives of Mari and Sippar. He collected the names of parents and
children whenever one or all of them bore an Amorite name. In 70 % of these cases, both the parent and the child had an Amorite name, whereas in 17 % of all cases, the parent had an Amorite
name, but the child had an Akkadian name. In the remaining 13 %, the names of either the parent
or the child were Akkadian, Hurrian, or Sumerian. The texts from Sippar that predate the reign of
Hammurabi indicated that in almost half of the cases both the parent and the child had Amorite
names. Akkadian-Amorite and Amorite-Akkadian parent-child combinations made up 30 % and
21 % of the total number of cases. In the later texts from Sippar, Amorite-Amorite name combinations were less common (11 %), whereas Amorite-Akkadian and Akkadian-Amorite parent-child
name combinations had risen to 53 % and 33 %, respectively (Streck 2004: 342).
7.3 Determining the nature of the Amorite identity
The previous sections have presented historical evidence for the presence of Amorites in greater Mesopotamia. Due to the distribution of the data, this review has emphasized southern
Mesopotamia over the north, and this bias will be addressed below. The previous sections have
shown that the label Amorite could be applied in a number of ways, and that it may have changed
its meaning over time. It is therefore important to discuss how the term Amorite may be interpreted. Previous researchers argued that the Amorites were nomadic pastoralists, possibly in the
process of becoming sedentary (Buccellati 1966; Kenyon 1966; Lönnqvist 2008), that they represented a rural population, to be juxtaposed against an Akkadian urban population (Buccellati 1990:
102), or that they came from northern Mesopotamia, and were already well-accustomed to urban
life (Porter 2007). However, the often proposed direct equation of Amorites with (semi-)nomadic
pastoralists, or a more general rural class, is not supported by the data presented above. Instead,
many researchers have described the Amorite identity as an ethnic label, allowing for a more
fuzzy distinction between pastoralists and agriculturalists, or nomads and city-dwellers (Kamp and
Yoffee 1980). However, this labelling is only useful if there is a shared understanding of the term
ethnicity. Therefore, the application of ethnicity to archaeological case-studies and the Amorite
case in particular will be discussed below.
7.3.1 Understanding ethnicity
There has been considerable debate on the nature of ethnicity, and on the application of ethnicity
in archaeology (e.g. Barth 1969a; Bentley 1987; Emberling 1997; Jones 1997; Stone 2003; Bahrani
2006), severely limiting the usefulness of the concept of ethnicity in the debate on the Amorites in
Mesopotamia. This debate suffered from (a) disagreement on a definition of ethnicity, (b) a historical burden resulting from an abusive association of ethnicity, race, and archaeology by modern ideologies and states, and (c) the inability to identify archaeological correlates of ethnicity. Principally,
there are two approaches to ethnicity: the primordialist perspective that explains ethnicity through
habitus, arguing that ethnicity is part of the unconscious values shared by a group; and the instrumentalist perspective that views ethnicity as an active identity that is maintained by an individual or
group in the interaction with others (Stone 2003: 32). The second approach is inspired by the work
of Barth (1969a), and has emerged as the dominant paradigm in the study of ethnicity (Emberling
1997: 295).97
97 Jones recently attempted to reconcile the two approaches by arguing that ‘ethnic identification involves an objectification of cultural practices (which otherwise constitute subliminal modes of behaviour) in the recognition and signification of difference in opposition to others’ (1997: 128).
129
The implications of recent anthropological research for the concept of ethnicity have not yet
been fully considered in all studies of ethnicity in the ancient Near East. For example, van Driel’s
recent evaluation of the concept of ethnicity (2005) is a large step backwards in this discussion,
completely neglecting Kamp and Yoffee’s (1980) much more thoughtful consideration of the subject. According to van Driel, ethnicity and its signalling primarily relate to problems caused by
conflicting land-use strategies (2005: 3). He further assumed that ethnic change occurred through
contact, Landnahme, and acculturation (van Driel 2005: 3-4). Thus, writing about the Amorites, van
Driel suggested that at the end of the Ur III period a Landnahme had occurred whereby land was
given to military colonists (2005: 7). Through this argumentation, ethnic change is reduced to a
process of migration and replacement of one social group by another, without consideration of
the complex interactions between ethnic groups that are documented for the more recent past.
Furthermore, the supposed Amorite take-over of power only seems to have occurred effortlessly
because there is a relative lack of historical evidence on exactly the period during which this takeover allegedly took place, that is, the period immediately following upon the Ur III period.
Fundamental for the modern conception of ethnicity has been Barth’s observation that ethnic
identity is not only acknowledged by others but also self-ascribed (1969a: 11). Within this approach
to ethnicity, it is not important to define an ethnic culture-core, but to define how, why, and when
ethnic groups distinguish themselves from other groups. This is logical, since any social division
can only be meaningfully articulated if both its members and any outsiders have a clear understanding of where and how the boundary or difference is drawn. Thus, according to Emberling
(1997: 304), ‘an ethnic group is most essentially a group whose members view themselves as having common ancestry, therefore as being kin. As kin units larger than any others, they must include
members of more than one lineage or extended family. Members of an ethnic group usually possess some common language. Ethnic groups often are unified by constructions of their past, by
perception of injustice in the past or in the present, and often by hopes of a future reunification.’
In this perspective, ethnicity is separated from other social categories including race, language, and
culture. Ethnic identity is the result of the dynamic social process between the actors themselves.
Rather than looking for what constitutes and is included in an ethnic identity, the differences between ethnicities should be the focus of attention. In archaeology, this translates into identifying
differences in material culture that are meaningful for the expression of ethnic differences (Stone
2003: 43). However, the situational nature of ethnic expression, that is, whether or not an individual or group decides to signal its ethnic identity, and through which channels that signalling is
achieved, makes it impossible for archaeologists to clearly identify ethnicity from archaeological or
historical records. No clear one-to-one relationships exist between ethnicity and particular cultural
traits (Stone 2003: 43). Rather, depending on the context, different features can be used to signal
ethnicity. Additionally, ethnicity is usually expressed through a combination of features (Horowitz
1985: 47). Because ethnicity is one of many identities an individual or group can assume, it must
further be considered whether particular features signal ethnicity at all, or not some other aspect
of identity (Emberling 1997: 299, 313; Stone 2003: 43).
Archaeologists thus face specific problems when attempting to identify the emergence and expression of ethnic identity. This is not only the case when ethnicity must be reconstructed from
material culture alone, but also when historical records naming groups and documenting interactions between them are available. For example, there has been a long-standing debate on how
the occurrence of non-Sumerian words in the earliest texts and the replacement of Sumerian by
Akkadian as the most important written language should be interpreted. This debate has often
been phrased in terms of migration. However, languages cannot be equated with ethnic identities, and other cultural and/or social processes may be invoked to explain changes in scribal traditions, which is the actual change observed by archaeologists and philologists (Yoffee 1995: 290;
Emberling 1997: 314). As regards the reconstruction of ethnic identities from material culture, this
can be most fruitfully pursued by using as many lines of evidence as possible. More specifically,
important social boundaries will be more strongly marked than less important ones (Emberling
1997: 318; Stone 2003: 43), although this correlation does not indicate if in a given context ethnicity is an important social structure.
130
pre-Ur III
Ur III
early second millennium
self-identification
ascribed identity
shared traits
-
+/-
-
-
+
-
+/-
+/-
+/-
Table 7.2: Comparison of available evidence for ethnic properties of Amorite identity through time (+: strong evidence;
+/-: some evidence; -: very little/no evidence).
There are numerous examples of ethnic change, that is, cases where an individual or group
changes its ethnic identity, and adopts a new identity. Because ethnicity is a social category, the
mechanisms through which it is changed are also of a social nature, that is, they involve the acceptance or adoption of new socio-cultural processes and/or ideas. Ethnic change first and foremost
assumes the acceptance of newcomers by members of the receiving ethnic group, and it assumes
that there are socio-cultural processes through which inclusion into the group can be achieved. If
such processes are in place, several motivations can lead to individuals or groups changing their
ethnic identities. On a very general level, such motivations entail a clear economical, cultural, or
social advantage of some sort to the group (Barth 1969a: 22; Abruzzi 1982: 21). Barth (1969b)
for example clearly showed in his analysis of Pathan ethnic identity that Pathans assumed new
identities when the situation forced them to do so, or when it was advantageous to them, despite a
clear reluctance to abandon the socio-cultural concepts associated with being Pathan. This ethnic
change was, however, only possible because the receiving group also saw advantages in increasing
their numbers, and in adopting new members into their social system (cf. Horowitz 1985: 48).
Having reviewed some important aspects of ethnicity as identity, it is now possible to discuss
whether or not these aspects apply to what is known about the Amorites. In other words, can the
Amorite identity indeed be considered an ethnic identity? Table 7.2 indicates the relative strength
of the evidence for ethnic properties of the Amorite identity for the three chronological periods
discussed earlier. As is clear from this table, the overall evidence is not very strong, at first sight
suggesting that the Amorite identity can only be considered an ethnicity during the Old Babylonian
period. However, there are reasons to believe that the weak ethnic markers of the Amorite identity
in the earlier periods are only partly due to the nature of the evidence, and that the Amorite identity may also be considered an ethnicity in those periods for which insufficient evidence is available.
Below, it will be argued that the absence of these markers can to a substantial degree be attributed
to the cultural and environmental contexts in which the Amorite identity developed.
7.3.2 The development of the Amorite identity
The development of the Amorite identity can be traced from c. 2500 bc, when it was first mentioned in historical sources. It is important to realize, however, that this date does not necessarily signify the emergence of that identity, but rather its appearance in the historical record. This
appearance indicates either that the writer(s) had become aware of the existence of an Amorite
identity, or that they considered that identity to be important enough to be mentioned. The scattered references to Amorites before the Ur III period are difficult to assess. Whereas the Amorites
fought by Naram-Sin were part of the army of Amar-Girid and may have been mercenaries, the
context of the conflict between Šar-kali-šarri and the Amorites was less clear. The Amorites may
have been the main opponents, but the reason for the battle remains unknown. On the other hand,
the earlier references to Amorites from Shuruppak and Lagash illustrate that already at that time
Amorites were integrated in Mesopotamian urban society. More importantly, they are mentioned
in very different contexts, that is, in a toponym, as recipients of barley rations, and as shepherds.
Although it is dangerous to rely on one or two textual references, these references nevertheless
suggest, together with what is known from the Ur III period, that Amorites did not enter lower
Mesopotamia as foreigners at the end of the third millennium bc, but were there well before that
time.
During the Ur III period, the historical sources point to two fundamentally different perceptions of Amorite individuals and groups (Buccellati 1966: 336). Administrative documents clearly
illustrate the high degree of integration of Amorite individuals in all levels of Ur III society.
131
More interestingly, Amorites could reach high positions in the state bureaucracy, as demonstrated
by Išbi-Erra, the later ruler of Isin. This observation contrasts the perception of Amorites as
emerges from the literary sources and the Royal Correspondence of Ur. These sources focus on
the unnatural and non-human aspects of the Amorite identity, thereby creating a clear dichotomy
between friends, that is, the Ur III state and its allies, and foes, that is, Amorite outsiders, that only
partially reflects reality. The official documents are generally concerned with the usually hostile
interactions between the state and Amorite groups. It seems therefore logical to assume that these
documents purposefully created an image of these Amorite groups as non-human enemies as a
kind of state propaganda (cf. Meijer forthcoming). The fact that the Ur III state considered these
groups as enemies derived not from their Amorite ethnicity per se, but from their raids into land
that was controlled or claimed by the state. Thus, some Amorite groups acted in a way that the Ur
III state deemed hostile, and it is specifically against these groups that the state employed its antiAmorite rhetoric, rather than against Amorites in general.
It is important to note in this context that the term Tidnum in Ur III sources occurred frequently as a synonym for Martu. Tidnum was mentioned in texts from Ebla and on Gudea Statue
B, but only as a geographical place name. The Tidnum in the Ebla texts cannot be precisely located, but the Tid(a)num in the ‘mountain range of the Amorites’ mentioned by Gudea should be
located in the Jebel Bishri. This location is based on Akkadian sources that clearly demonstrate
that the ‘mountain range of the Amorites’ was another name for Bašar. However, in the Royal
Correspondence of Ur, Tidnum is used to designate a group of people, and more specifically as
the people against which the ‘wall of the land’ was built. The Šu-sin text, in which he describes his
victory over the Amorites, suggests that the Tidnum were one among several Amorite subgroups,
thus again reinforcing the view that not all Amorites were hostile toward the Ur III state, but only
specific subgroups. The combined evidence on the location of the ‘wall of the land’, as well as
some references to Tidnum in combination with geographical places that can be located in the
northeastern Zagros suggests that the Tidnum probably lived north and northeast of the Ur III
state (Sallaberger 2007: 448-9). Thus, Tidnum as a place name can be located in the Jebel Bishri,
whereas the group identified as such must be sought in the Zagros. The references to Tidnum in
Ebla and possibly Ur cannot be interpreted, but may point to additional regions where the name
Tidnum was known. Concluding, the use of the same name for different regions, or for Amorite
people living there, may be an early reflection of the mirror toponymy that is clearly attested for
the early second millennium bc (see below).
The early second millennium bc represents a third stage in the development of the Amorite
identity. Administrative texts no longer labelled individuals as Amorites, but Amorites were still
recognizable as such by their distinctive names. Furthermore, as Streck’s analysis has shown, there
was a proliferation of Amorite personal names across greater Mesopotamia and western Syria,
with Amorite names occurring as far south as Hazor. At the same time, royal inscriptions appeared
in which rulers from cities across greater Mesopotamia pointed out their Amorite ancestry, and
described themselves as being Amorites. These claims represented a fundamental change from
the royal historical sources of the Ur III period, which were generally concerned with portraying
Amorites as enemies of the state, and which did not consider them to be human at all. Apparently,
early second millennium rulers found it advantageous to emphasize their Amorite ancestry. Many
researchers have interpreted this shift as a result of some sort of Amorite population displacement at the end of the third millennium, leading to the replacement of the old urban elites by
new Amorite rulers (e.g. Kenyon 1966: 34; Klengel 1972: 44; Weiss et al. 1993: 1002; Charpin et al.
2004: 57-60; Buccellati 2008: 143). However, the historical and archaeological evidence does not
uniformly support this interpretation (Yoffee 1988: 50). Furthermore, this interpretation is based
on flawed analogies between Amorite-Ur III interactions and ethnohistorically documented nomad-sedentary interactions.
First, the physical passage of Amorites through the wider Near East in the early second millennium bc has been inferred from the occurrence of identical place names from western Syria to
northern and southern Mesopotamia. Charpin called this phenomenon a ‘toponymie en mirroir’
or mirror toponymy (Charpin 2003). This phenomenon is most clearly seen in letters that actually state that there are two places with the same name, but can also be inferred from place names
132
where a qualification is added, that is, whether the city is located near a river, or whether it is associated with a particular deity (Charpin 2003: 7-10). Charpin suggested four possible mechanisms
for the establishment of double place names: (a) chance, that is, place names with obvious meanings or roots like ‘encampment’ or ‘ruin’ are likely to occur multiple times without any further link
existing between them, (b) deportations whereby the deportees introduced new names into the
region to which they were relocated, (c) the place names were clearly meaningfully linked in some
way, as may possibly be inferred from the double occurrence of groups of place names between
region, and (d) the place names were established in the course of a migration of Amorite groups
across the Near East (Charpin 2003: 12-15). Whereas the first three mechanisms seem to have
operated only in a limited number of cases, Charpin argues that the fourth mechanism must have
been very important. The main argument in favour of migration is that none of the early second
millennium bc toponyms is found in third millennium texts, and that a systematic renaming would
have occurred, whereby ancient names were dropped in favour of new, proper Amorite names
(Charpin 2003).
There are several arguments against this interpretation. First, it is unclear whether many toponyms in northern Mesopotamia actually were of Amorite origin (Eidem 2000: 262). Secondly,
Porter (2009: 204-5) recently argued that, in the absence of information on when each place name
came into use, it is equally possible that the duplication of geographical names was intended to
reinforce a common social identity over large distances. In this perspective, the adoption of identical place names served to actively strengthen social ties between possibly mobile groups that
met only irregularly. In other words, the place names themselves became part of social structure.
Based on the historical evidence alone, neither the explanation favouring migration nor favouring
changes in social structures can be ruled out. At the same time, the mirror toponymy can in and of
itself no longer be used to infer migration, unless other evidence can be produced to support that
hypothesis. Thirdly, it should be noted that the mirror toponymy of the early second millennium
bc is primarily known from the Mari texts and contemporary archives elsewhere. The period covered by these archives, as well as smaller contemporary text corpuses, is exceptionally well-known
in terms of historical geography. The somewhat sudden occurrence of a mirror toponymy might
therefore at least partially be the result of an increase in the amount of information that is available for this period with respect to previous periods, and of an increase in the geographical range
that is covered by the Mari texts. Finally, the possible occurrence of a mirror toponymy prior to
and during the Ur III period suggests that the early second millennium bc can no longer be regarded as the only period during which this phenomenon occurred, thereby further weakening the
link between the mirror toponymy and a unique episode of large-scale migration during the early
second millennium bc.
A second argument against an Amorite migration comes from Streck’s analysis of the Amorite
onomasticon. As described earlier (Section 7.2.3), Streck has demonstrated that Amorite names
can appear together with Akkadian and Hurrian names within the same family, and that parents
with non-Amorite names could give their children Amorite names. in so much as ethnic affiliation
can be deduced from personal names only, these data indicate that ethnicity may have been a relatively fluid concept, and that naming conventions may have changed over one or two generations.
If there existed proper naming conventions, and if names were a defining trait of the Amorite ethnicity, this evidence does not support the replacement of previous populations by a new Amorite
population. This evidence rather suggests that becoming Amorite was a conscious choice that was
open to at least a part of Mesopotamian society.
The ethnic fluidity that is implied by the onomastic evidence is also apparent in a letter from
Mari from the high official Sammetar to Zimri-Lim, describing a group’s transition from the
Yaminite into the Sim’alite confederacy: ‘Another thing. Uranum and the elders of Dabiš came,
saying: “By extraction, we are among the Yahrurû, but never (as) yarrādum; also, in the back country, we have neither hibrum nor kadûm.” We are native (?) to the Yahrurû. Let us now come into the
midst of the Sim’alite(s) as (part of) the Nihadû (tribe), so that we may slay the (treaty) donkey’
(Durand 1992: 118; Fleming 2004: 63). In the remainder of the letter, Sammatar asks the elders
of Dabiš three times whether they want him to write to the king about this matter. In the final
paragraph, Sammatar requests the king’s permission to carry out the ritual that is necessary to turn
133
the Sim’alites into Yaminites. The Yahrurû are one of the five Yaminite li’mum, and the Nihadû are
a known Sim’alite gayum. This letter makes clear, then, that social mobility between the two tribal
confederacies was possible if there was consensus among all involved parties.
Thirdly, the idea that Amorites forcefully made forays into regions controlled by sedentary
states is a direct analogy of the situation in the Near East as it existed during the late Ottoman
period, and more particularly in the areas that today make up Jordan and Syria. If this analogy is
examined more closely, it becomes apparent that many aspects of nomadic pastoralist society during the Ottoman period were absent or different during the third and early second millennium bc
(cf. van der Steen 2009: 114). The Ottoman government paid Bedouin sheikhs to protect villages
from raids by other Bedouin. This system was, however, not very effective, and sedentary farming villages often had to pay tribute or khuwah to more than one Bedouin tribe (Lewis 1987: 12).
There is currently no textual evidence to suggest that a similar system functioned during the Ur
III or Old Babylonian periods. The success of Bedouin raids depended not only on the fact that
Ottoman protection of these villages was insufficient, but also on their ability to quickly disappear into the steppe when Ottoman soldiers did appear. Bedouin were able to do so because they
were nomadic camel-pastoralists. In this respect it should be mentioned that third/early second
millennium bc (semi-)nomadic groups were probably much less mobile than later groups because
horse and camel had not yet been introduced as riding animals (Khazanov 2009: 123). Khazanov
furthermore suggested that the absence of the camel may have severely limited the number of
people that could be sustained by steppe areas, simply because these areas could not be exploited
to the fullest by sheep/goat pastoralists. Thus, any third to early second millennium bc equivalents
of the Bedouin would be severely limited in their mobility, their numbers, and consequently their
military advantage against more sedentary enemies (Khazanov 2009: 123).
The Ur III did not end because of attacks by Amorites, but Amorite groups, as well as others,
were able to raid into Ur III territories because the state was already weakened by internal processes. Išbi-Erra’s rise to power is a case in point. According to the Royal Correspondence of Ur,
Išbi-Erra was an important official in the Ur III state before he became the ruler of Isin. Not only
does this case indicate that Amorites were well-integrated into Ur III society, but also that, at least
in the case of Isin, the emergence of Amorite rulers was not related to, or caused by, any population displacements. Išbi-Erra’s usurpation is more indicative of internal power struggles during the
final years of the Ur III state, in which Amorites actively participated (cf. Wilcke 1969: 15; Yoffee
1995: 297).
Finally, another striking aspect of the emergence of an Amorite identity is the fluidity with
which Amorite rulers embedded themselves in a pre-existing urban culture. Thus, in terms of material culture and the expression of kingship, Amorite rulers and supposedly Amorite towns were
not very different from what preceded them. Traditionally, this continuity has been interpreted
either as an attempt by Amorite rulers to imitate or adopt pre-existing structures to legitimize
Amorite rule, or as evidence for a complete absorption of the Amorite newcomers into the supposedly more sophisticated urbanized Mesopotamian cultural traditions (Buccellati 1966: 355-6).
The problem that is posed by this fluid transition was correctly realized by Porter, who therefore recently suggested that the Amorites were ‘part and parcel of the mainstream urban world’
(2007: 70). Instead of arguing how and why nomadic pastoralists could have overthrown the Ur
III state, only to be rapidly assimilated into the urbanized Mesopotamian culture, she suggested that the Amorites originated from northern Mesopotamia, where they were already familiar
with an urbanized environment. Thus, Porter explained the impact of the Amorites on southern
Mesopotamia primarily in terms of the heterarchical system of city-states that replaced the Ur III
state (Porter 2007: 107). Porter rightly drew attention to a number of important issues, some of
which were also emphasized by other researchers: (a) the possibility that the Amorites originated
not from marginal steppe areas but from more fertile regions in the northern Jezirah (Meijer 2000:
206), (b) the possibility that there may have been no assimilation of the Amorites into southern
Mesopotamian culture because they already possessed many similar cultural features (Kamp and
Yoffee 1980: 98-9; Yoffee 1988: 50), and (c) the fact that the end of the Ur III state represented
primarily a transition towards a scattered political landscape of independent city-states retaining
much of their previous socio-cultural traditions (Yoffee 1995: 297).
134
However, whereas Porter successfully disassociated herself from the traditional view of
Amorites as fulltime nomads and pastoralists, she, as well as many other researchers, failed to
completely move away from the idea that the emergence of an Amorite identity must have entailed
some sort of population displacement. As was argued above, neither the idea that there was a
large-scale Amorite population displacement, nor the idea that Amorite nomadic pastoralists constantly pressed into southern Mesopotamia from the more marginal steppe regions, fully explains
the available evidence. Especially the observation that at least some Amorites were obviously wellintegrated into southern Mesopotamian society already before the final centuries of the third millennium bc remains unexplained.
7.3.3 The Amorite identity as ethnicity
In order to address the issues that have been raised in the previous section, and more importantly
to explain the particular development of the Amorite identity throughout the third and early second millennium, the following model will be put forward as an alternative to previous explanations. Rather than focussing on population displacement as the primary cause of the emergence
of Amorite states in greater Mesopotamia during the early second millennium bc, this model emphasizes ethnicity as a flexible identity that can emerge and disappear without necessarily involving
population displacement.
The model that is presented here argues that Amorite is an ethnic identity that was at least
partly constructed around a shared past and ancestry. The use of the determinative Martu in Ur III
records, as well as the references to Amorite ancestors by early second millennium rulers, reflect a
general concern for maintaining an identity that is firmly rooted in a perceived or constructed past.
From this perspective, Shamshi-Adad’s and Hammurabi’s claims to the same ancestral roots need
not necessarily reflect real kinship. In fact, it is possible that these genealogies were built around
the social networks in which Shamshi-Adad, Hammurabi and other early second millennium rulers participated, rather than that these networks emerged out of true historical kinship ties. This
loose use of kinship to legitimize and strengthen existing social relations has been observed among
Rwala Bedouin. They deliberately lost track of their genealogies after four or five generations in
order to maintain a flexibility of membership to particular lineages that could be modified when
necessary (Barfield 1993: 74-6). The system of constructed kinship among the Rwala was characterized by the persistence of a limited number of named ancestors, even though these should
have changed if the system reflected true kinship (Barfield 1993: 75). The persistence of the term
Tidnum over several centuries and its slightly different meanings, that is, as tribal and ancestral
name, might reflect a similar intentionally fuzzy system in which true kinship was of secondary
importance to the maintenance of a common ethnic identity. 98
The evidence for a separate Amorite language possibly further strengthens the interpretation
of the Amorite identity as a distinct ethnicity. However, it is difficult to determine whether language was considered a defining trait of the Amorite identity. The frequent occurrence at Sippar
and Mari of Amorite names in families previously dominated by Akkadian names might suggest at
least a minimal knowledge of or familiarity with Amorite. The continuing dominance of Akkadian
as written language does not necessarily reflect the linguistic composition of greater Mesopotamia,
as is clear from the references to people who spoke multiple languages. It would in this regard be
interesting to know the degree to which speakers of Amorite and Akkadian could understand each
other. Although Buccellati has suggested that the languages may initially have been ‘socio-lects’
with a significant degree of overlap (1992: 98), the mentioning of an Amorite interpreter in the Ur
III texts (Section 7.2.2) suggests that this mutual understanding was rather low.
The strong association with pastoralism evident in texts from the Ur III period and the early
second millennium bc represents an additional common feature of the Amorite identity. This association does not necessarily mean that all Amorites were pastoralists, which they clearly were
not, but rather that many Amorites were pastoralists, and, more importantly, that many pastoral98 The names, or parts thereof, of several Sim’alite gayum during the reign of Zimri-Lim were structured as personal
names (Durand 2004b: 183). This could indicate a similar concern with ancestry as is evident in the genealogies of
Hammurabi and Shamshi-Adad.
135
ists were Amorites. It is here suggested that practising pastoralism may have represented an ideal
life-way by which ultimately all Amorites were defined, including those that were not pastoralists
themselves. In other words, being pastoralist was at the core of being Amorite, but being Amorite
did not require being pastoralist. It was this aspect of the Amorite ethnic identity that probably set
it apart from other contemporary ethnicities.
This reconstruction of the Amorite identity as representing an ideal pastoralist life-way has
parallels elsewhere. In a study on Maasai identity in East Africa, Galaty (1982) found that the
Maasai defined their own and others’ identities on three levels nested within each other. At the
most universal, mythological level, Maasai distinguished between themselves, being ‘people of cattle’, and hunters and agriculturalists. Myths explained why hunters and agriculturalists were nonMaasai by definition. At the intermediate level, representing a division of the Maasai identity itself,
there were true Maasai or pastoralists, Maasai hunters, and Maasai agriculturalists. Finally, the third
level represented a division of labour within true pastoralist groups. At this level, there were true
pastoralists or Maasai, as opposed to the class of craftsmen, and the class of diviners. Even though
at each level the strict meaning of Maasai was pastoralist, the Maasai identity could at times include
other persons and groups as well. Even at the mythological level, some non-Maasai could be classified as Maasai with respect to outsiders. There thus existed clear distinctions between how the
Maasai identity was defined, and how it was applied to different groups or persons. Nevertheless,
the pastoralist ideal of Maasai identity was present at each level of identification (Galaty 1982).
Similar distinctions between ancestry and practiced subsistence strategies also played an important role in more recent Near Eastern agriculturalist/pastoralist communities. In the Balikh Valley,
Syria, there was at least from the nineteenth century ad onward a clear distinction between ‘Arab,
hadar, and badu (Lewis 1988: 686). ‘Arab was used for people who were semi-nomadic at that time
or had been in the past, and who were of tribal descent. Hadar was used for people who had always
been sedentary, and who could not claim a tribal descent. Badu indicated people who had been
fully nomadic camel-pastoralists. Even though these distinctions had become virtually meaningless
from an economic perspective by the time of Lewis’ ethnographic research, the maintenance of
these distinctions indicated the persistence of the ideals that were associated with each identity.
In this sense, then, Amorites may perhaps be best understood as ‘people of sheep and goat’,
thereby referring to an aspect of Amorite identity through which all Amorites identified themselves without actually always being involved in it. The discrepancy between ideal and reality is
perhaps best reflected in the apparent inconsistency that can be found in the literary composition
of The marriage of Martu, noted by Porter (2007: 108). On the one hand, this composition described
the god Martu as living in a tent, eating raw meat and so forth (see Section 7.2.2), whereas on the
other hand it was said that he lived in the city of Inab. It is clear, then, that in this text the Amorite
ideal is distorted as part of the story being told, but that the composition may have nevertheless
reflected the complexities of the interaction between Amorites and others, and among Amorites
themselves. King Zimri-Lim of Mari called himself ‘king of Mari and the land of the hana’, thereby further emphasizing the Amorite pastoralist ideal.
If it is accepted that people could adopt the Amorite ethnic identity, there must have been clear
advantages to this choice. Given the fact that the strong association with pastoralism was a defining and integral aspect of Amorite identity, it is likely that any advantages resulted from precisely
this association. The strong association with, but not exclusive focus on, pastoralism makes it likely
that the Amorite identity functioned as a social expression of an exchange network between pastoralist and agriculturalist groups. The occurrence of duplicate place names across Mesopotamia
may also point to a consciously maintained shared identity. The shared Amorite identity positively
influenced the emergence and maintenance of intra-ethnic exchange networks, giving Amorites
an adaptive advantage that allowed them to respond adequately to any crisis situation they might
encounter.
136
7.4 Conclusions
The present chapter has reviewed textual evidence for the presence of Amorites in Mesopotamia.
Amorites were mentioned in historical records from c. 2500 bc onwards, with a strong emphasis
on the Ur III period, as well as the early second millennium bc. This evidence indicates that the
perception of Amorites by those who wrote about them changed over time. Whereas the midthird millennium texts are difficult to assess, the Ur III texts clearly show a discrepancy between
the perception of Amorites in literary sources and administrative documents. The literary sources
provide a stereotypical view of Amorites as being uncivilized and hostile toward the Ur III state,
whereas the administrative texts indicate that Amorites were well-integrated into Ur III society.
The early-second millennium material shows that Amorites could be found in all levels of society,
and that many rulers across greater Mesopotamia were proud of their Amorite heritage. The Mari
archives furthermore show that Amorite states were not organized on an agriculturalist-pastoralist
dichotomy, as has been previously emphasized, but that these states were based on tribal identities that encompassed both economies. This evidence suggests that the incorporation of both the
pastoral as well as the agricultural economy may have been the defining aspect of Amorite identity. The most important conclusion must be that the Amorite ethnicity emerged as an identity
that can be interpreted as signifying a social network extending over large distances. The spatial
distribution of the literary sources points to a strong focus on lower Mesopotamia, as well as the
Syrian Euphrates and Khabur basins. Thus, the Amorite identity not only transcended boundaries
between subsistence systems, but environmental boundaries as well. These characteristics gave
Amorites a behavioural flexibility when responding to environmental stress. The development of
the Amorite identity with respect to environmental change, and with respect to the hypotheses that
were formulated in Section 3.5.3, will be discussed in the following chapter (Section 8.3).
137
8 Social responses to environmental change
8.1 Introduction
The palaeoenvironmental data presented in Section 2.3 suggest that northern Mesopotamia experienced increasing aridity from c. 2500 bc onwards. Although the effects may have been quite
localized at that time, the data suggest that aridity increased toward the end of the third millennium bc. For that period, all local northern Mesopotamian palaeoclimate records, that is, Kazane
Höyük, Göbekli Tepe, Wadi Avedji, Wadi Jaghjagh, and Tell Leilan, indicated a shift towards greater aridity (Fig. 2.7F–J). These records also suggest that this shift was not a short-lived event, but
initiated a period with a structurally drier climate than existed during the early third millennium.
Furthermore, osteological and botanical remains suggest that humans and plants may have experienced nutritional stress and water shortage, supporting the hypothesis that the late third millennium bc arid transition indeed affected human communities and led to more droughts. This climate
scenario, then, is the background against which the changes in social structures that were reconstructed in Chapters 5 through 7 should be evaluated. The reconstructed developments for each
survey region will be analyzed and discussed in light of the expected interaction patterns. Then,
the textual evidence presented in Chapter 7 will be evaluated and compared with the development
of the pastoral economy that was reconstructed in Chapter 6. Finally, this chapter hopes to place
the reconstructions in a wider regional context, and to assess the degree to which responses across
northern Mesopotamia were similar, and how local responses related to other developments.
8.2 The development of sedentary-sedentary relations
In Section 3.5.1, it was argued that a decrease in the carrying capacity within a region would lead
to increased competition between groups relying on the same resources. For groups relying on
subsistence farming, this competition would likely focus on good quality arable land and on water. Increasing competition over these resources would likely result in the development of social
mechanisms regulating access to and use rights of these resources. Based on the theoretical framework outlined in Section 3.5, a convex rank-size distribution is especially expected during periods
of resource stress. Chapter 5 described the selection of three case-studies in which this hypothesis was tested. These case-studies were the region covered by the Euphrates-Birecik Dam Survey
(B-EDS) in Turkey, the Syrian part of the Balikh Valley that was covered by the Balikh Survey
(BS), and the region covered by the North Jazira Survey (NJS) in Iraq. Diachronic developments
in site size, as well as rank-size distributions, were calculated as proxies of inter-site competition.
Furthermore, regional population trends were calculated to assess the degree to which northern
Mesopotamia was subject to de-urbanization and de-population during the late third–early second
millennium bc.
The B-EDS region was characterized by a low degree of urbanization and a low population
density throughout the third and early second millennium bc. Population generally increased, with
a minor decrease between 2100 and 1700 bc. Average settlement size steadily increased throughout
this period. Above-average population growth is recorded for the Late EBA/Early MBA period,
and the end of the Late MBA (Fig. 5.10A). The B-EDS rank-size plots were extremely primate
and no significant diachronic change could be observed due to the overlapping error ranges (Fig.
5.11). The BS demographic curve showed alternating periods of population growth and decline
during the third and early second millennium bc. The settled population grew during periods VIab,
VIIa, and VIIb, and declined during periods VIcd and VIIc (Fig. 5.10B). At the same time, average settlement population peaked around 2000 bc at the VIcd–VIIa transition, only to go into a
139
steep decline toward 1600 bc. Whereas the population growth rate during period VIab fell within
the range of natural growth, the period VIIa–b growth rate exceeded the natural growth rate (Fig.
5.10B). The rank-size analysis showed a probably convex size distribution for the VIIa–VIIb and
VIIb–VIIc transitions, but the large confidence zones for especially the 2400 bc and 2000 bc plots
impede a diachronic reconstruction (Fig. 5.12). Finally, the NJS showed a steep population growth
during the early third millennium, followed by a period of high demographic stability that ended in
a marked population decline from 1750 bc onwards (Fig. 5.10C). Average settlement size paralleled
this development. Population growth exceeded natural growth during the early third millennium,
but remained within the natural range for the entire second half of the third millennium, as well
as the first quarter of the second millennium bc (Fig. 5.10C). The NJS plots were slightly primate,
with a possibly primo-convex distribution observed for 3000 bc. However, the large confidence
zone for each reconstructed rank-size plot suggests that these distributions must be treated with
caution and that conclusions drawn from them are not very reliable (Fig. 5.13).
From these data, two interesting observations can be made with respect to the scenario of environmental change: all regions experienced population growth at some point in time during the
late third–early second millennium bc when aridity increased; and a convex distribution is only
observed in the BS region at the VIIa–VIIb and VIIb–VIIc transitions at c. 1800 and 1700 bc, respectively, whereas in the B-EDS and NJS regions, there is no evidence at all for a convex size distribution. These observations directly allow the conclusion to be drawn that no direct correlation
existed between increasing aridity and population decline in settled communities. This conclusion
is supported by the evidence from other surveys (Section 5.4.1). In each survey region, the number
of sites and the aggregate site size increased and declined considerably during the late third–early
second millennium bc, but no apparent pattern and certainly no complete synchronicity can be
directly inferred from these developments. Furthermore, the absence of convex settlement distributions, except for a short period in the Balikh Valley, suggests that climate change may have had
a relatively minor impact on the organization of the settlement systems that were analyzed in this
study. The following sections will analyze the developments in each survey region in more detail,
as well as discuss the role of climate change in the reconstructed developments.
8.2.1 The Upper Euphrates: the emergence of Carchemish
The number of settlements in the B-EDS region is low throughout the third and early second millennium bc. The possible slight change toward a more primate rank-size distribution could reflect
the growing importance of Carchemish toward the late second millennium, as is possibly also reflected in the site’s increased size. Clearly, these settlement developments fall outside the expected
convex settlement pattern resulting from inter-site competition and growing resource stress.
It was already suggested that the area covered by the B-EDS probably represented the northern
half of the Carchemish sector, whereas the southern half is located to the south of the modern
Turkish–Syrian border (Section 5.3.2). This would suggest, then, that the primate nature of the BEDS rank-size plots resulted from the fact that the total settlement system had not been correctly
identified. Nevertheless, it should be realized that the areas north and south of Carchemish were
very distinct in terms of environment and settlement patterns. The area north of Carchemish,
that is, the B-EDS region, was characterized by relatively small settlements that were located close
together, whereas the area south of Carchemish had a much more dispersed settlement system
(Peltenburg 2007a: 17–18). Secondly, Carchemish is located in the transitional zone between the
stable dry-farming wheat-based agricultural production system of the northern part of the valley,
and the barley-based system in the southern part of the valley (see Fig. 2.5). Finally, Carchemish
lies in the transitional zone between the dense oak woodlands of the Taurus proper, and the woodland steppe that characterized much of the Upper Syrian Euphrates (Peltenburg 2007a: fig. 1.3).
Given the fact that the B-EDS area falls well within the zone of dry-farming, it may be questioned whether the region ever experienced a sufficiently severe drought as to permanently affect
the groups living in the region. This situation would explain why the expected settlement pattern
did not emerge in this region as a response to increasing aridity. Nevertheless, a correlation between climate change and the settlement development in the B-EDS region may be tentatively
140
Social responses to environmental change
suggested. Large sites in this region may have played an important role in north–south interaction.
The location of this region between different environmental zones would suggest that Carchemish
functioned as a gateway city between these regions (see Burghardt 1971).
Summarizing, the development of the settlement system in the B-EDS region seems to have
been relatively unaffected by climate change during the late third millennium. The emergence of
Carchemish as an important gateway city between the northern and southern environmental and
cultural zones may potentially be linked to increasing aridity, as communities in the more marginal
regions sought to establish social relations with the north.
8.2.2 The Balikh Valley: cycles of competition
Close inspection of the data from the BS region suggests that population growth, rather than climate change may have been the driving force behind increasing resource competition. During the
third and early second millennium bc, the Balikh Valley experienced two periods of population
growth, resulting in peak populations at 2400 and 1700 bc. Wilkinson suggested that the recent
water discharge of the Balikh, before the introduction of large-scale motorized irrigation, was
potentially sufficient to irrigate agricultural fields sustaining a population of 7000–30000 persons
(1998: 80–1).99 It can safely be assumed that the late third–early second millennium discharge was
at most the same as this amount. The reconstructed population between 3000 and 1600 bc continuously exceeds the lower end of this range. The population peak of 2400 bc, and particularly that
of 1700 bc, approached the upper limit of this range, suggesting that agricultural production may
have been subject to stress and periodic failure. It is exactly during these two periods of maximum
population that the strongest evidence for competition between communities can be found. 100
Towards 2400 bc, the inter-community social tension building up in the Balikh Valley was
reflected in the construction of city walls documented for the middle and late third millennium
bc. Hammam et-Turkman (BS 175) was walled around 2350 bc (Thissen 1989: 197). Around the
same time, city walls were built at Jidle (BS 275) and possibly Sahlan (BS 247) (Mallowan 1946:
134, 138), Kazane Höyük in the Harran area to the north of the BS region (Rosen 1997: 397), and
the city wall of Tell Bi’a (BS 1) was renewed (Miglus and Strommenger 2002: 21). These building activities reflect a real concern for safety and occurred at a time when pressure on resources
was growing. In this system, no site was able to assert control over the entire region to extract
resources for its own benefit, a system that may have functioned for some large third millennium
centres elsewhere in northern Mesopotamia (Wilkinson 1994). One reason why this type of settlement hierarchy did not appear in the Balikh Valley may have been that not land but water was the
restricting resource.
During the early second millennium, there is direct evidence for competition over water in the
letter to Shamshi-Adad from the inhabitants of Tuttul/Tell Bi’a. They complained that the city of
Zalpah, located upstream along the Balikh and possibly to be identified with Hammam et-Turkman
(van Loon and Meijer 1988: xxvi), cut off the waters of the Balikh, thereby depriving Tuttul of vital irrigation water (Dossin 1974; Villard 1987). Given the fact that the immediate surroundings of
Tell Bi’a are far more easily irrigated by using water from the Balikh than the Euphrates (Schirmer
1987: 71), this complaint seems fully justified if there actually was water shortage. Tell Bi’a may not
have been the only settlement to suffer from reduced river discharge, and the resulting competition
over resources may have led to settlement nucleation and the formation of distinct territories, as
evidenced in the convex rank-size distributions for the period 1800–1700 bc.
99 It remains unclear whether irrigation was actually practised in the Balikh Valley during the third and early second
millennium bc. Clearly identifiable ancient irrigation canals in the valley dated to the Hellenistic period and later
(Wilkinson 1998: 69). Botanical samples from Hammam et-Turkman period VI and VII contexts contained too few
weed species indicative of a wet environment, and by inference, of irrigation, to posit the practice of irrigation (van
Zeist et al. 1988: 713).
100 The present reconstruction of the Balikh Valley demographic developments supersedes, and slightly differs from, the
analysis presented in Wossink forthcoming. The differences mainly derive from the inclusion of the results of the
rank-size analysis, necessitating a modification of the scenario of human-environment interaction that was offered
earlier.
141
This scenario differs considerably from Curvers’ (1991: 215) interpretation of the settlement
system in the Balikh Valley during the Early and Middle Bronze Age. Curvers’ model was primarily
driven by a cycle of waxing and waning southern Mesopotamian influence on northern city-states.
According to Curvers, the mid-third millennium bc population peak, which he called ‘process of
urbanization’, resulted from the ‘catalytic effect of Uruk expansion towards Syria and Anatolia’,
and was associated with the emergence of a ranked society in the Balikh Valley. During the late
third millennium bc, the complexity of this society increased due to influences from southern
Mesopotamia. At the end of the third millennium, this system collapsed due to internal social and
political stress. The second ‘process of urbanization’ was initiated by local urban rulers who had
a strong pastoral background and who pursued a policy of (re)settlement and sedentarization.
These rulers sought to establish their power base by associating themselves with powerful foreign
dynasties. Eventually, after 1700 bc, this system collapsed because these foreign powers withdrew
their support from the local rulers in the Balikh Valley. Curvers emphasized the influence of external, primarily southern Mesopotamian, political entities, that is, the Uruk phenomenon, Ebla,
the Akkadian and Ur III states, and Mari., This study however focuses on local processes as the
primary causes of the observed socio-cultural changes. Local human-environment interaction are
emphasized over the still debated impact of far-off political entities, and emphasizes political,
cultural, and economic continuity in the Balikh Valley. In this way, there is no need to hypothesize
significant southern Mesopotamian influence on the Jezirah, for which archaeological and textual
evidence remains ambiguous (see Section 2.5).
Summarizing, it can be argued that in this scenario, both the mid-third millennium bc as well
as the early second millennium bc cycles of increasing competition were primarily triggered by
population growth, resulting in higher population pressure on available resources. The onset of
drier climatic conditions during the second half of the third millennium, resulting in reduced river
discharge and less reliable rainfall, may have strained agricultural production and led to a stronger
reliance on irrigation, but was otherwise not a driving force behind the settlement developments
that were reconstructed for the Balikh Valley for the period 3000–1500 bc.
8.2.3 The northeastern Jezirah: centralization and exchange
The reconstructed settlement data from the NJS region are difficult to interpret. No clear convex
settlement distribution can be discerned due to the large error margin of the NJS rank-size plot.
In fact, all plots demonstrate a slight tendency toward a primate or primo-convex distribution. It
has been suggested that the NJS region was most likely to represent an area that did not coincide
with an entire settlement system, as a result of which the rank-size plots would be flawed (Section
5.3). The proximity of large sites outside the survey area would suggest on the one hand that the
NJS region may harbour parts of other settlement systems, and on the other hand that the system
that can be associated with Tell al-Hawa was not entirely included in the NJS region.
Despite these uncertainties, a convincing case has been made for the existence of a mid–
late third millennium hierarchical settlement system centred on Tell al-Hawa (Wilkinson 1994;
Wilkinson and Tucker 1995). The model was based on the survey evidence from the NJS, consisting not only of the settlement record, but also of the hollow ways and extensive sherd scatters
around major sites. According to this model, the hierarchical system was centred on a large central
site. This site’s population was so large that in a dry year with low agricultural production, its own
catchment area, hypothesized to be an area around the site with a radius of up to 5 km, could no
longer sustain the population, and surplus had to be imported from surrounding sites. Through
the extraction of agricultural surplus from surrounding sites, which remained below their site
catchment’s carrying capacity, the central site was able to grow beyond its own catchment area’s
carrying capacity. However, whereas the satellites around Tell al-Hawa were abandoned, some settlements around the second-rank sites remained occupied, suggesting a slightly different land-use
system for the central sites and the second-rank sites. This late third millennium system probably
emerged out of a Late Chalcolithic settlement landscape characterized by small independent villages, through an early third millennium system characterized by central sites surrounded by satellites within a 5 km zone (Wilkinson 1994; Wilkinson and Tucker 1995).
142
This hierarchical settlement system relied on strong interaction between the central site and
the secondary settlements. Grain deliveries must have been recorded, and there must have been
regular traffic between the settlements within the system, especially in times of local labour shortages during harvest (Wilkinson 1994: 503). Some evidence for these contacts came from the area
of Titrish Höyük along the Turkish Euphrates, where a hierarchical settlement system similar to
that of Tell al-Hawa was argued to have existed. At Titrish Höyük, a particular type of large storage jar was abundantly present at the central site, but largely absent at the secondary sites, and
this difference was speculatively argued to represent a flow of grain toward the central site. This
pattern could also be identified in the NJS area, but not as strong as in the area of Titrish Höyük
(Wilkinson 1994: 494–5). The slight primate tendency observed in the rank-size plots of the NJS
region could potentially also point to a vertically integrated settlement system, although this would
require a better understanding of the limits of the Tell al-Hawa settlement system as a whole (see
Section 5.3.2).
Apart from the emergence of large, centralized sites, the model also aimed to explain the
evidence for agricultural intensification. Shortening of fallow and applying animal manure to the
fields would have been the primary methods of intensification. Whereas fallow-shortening cannot be proven from archaeological remains, there is circumstantial evidence for animal manuring.
Many mid–late third millennium bc urban centres in the Upper Khabur area and the northeastern
Jezirah are surrounded by hollow ways, supposed to have come into being through continued traffic by persons and animals (van Liere and Lauffray 1954; Wilkinson 1993; Ur 2003; Rutishauser
2008).101 Around many of these sites, extensive sherd scatters can be found that extend as far away
from the site as many hollow ways. Wilkinson suggested that these sherd scatters represent the
material remains of settlement refuse being applied as manure to agricultural fields (1989), and
that the hollow ways may have indicated the limit of the zone of intensive agricultural production.
Given that the sherd scatters occur around sites that reached their largest size during the mid–late
third millennium, it seems reasonable to assume that manuring was adopted as a method to achieve
higher agricultural production.102 Calculations of mean annual production per hectare, combined
with reconstructed population levels for these large sites, have suggested that these populations
could indeed only be sustained under an agricultural regime combining short or no fallow periods
with manuring and/or irrigation (Wilkinson 1994, 2004; Altaweel 2008).
Looking more closely at this model of land-use around urban sites, it becomes apparent that
the model combines two developments that are not necessarily related. First, the model proposes
how a central site can attain a size beyond that which can be sustained by its immediate catchment
area through extracting surplus from neighbouring sites, and secondly, the model proposes how
the overall population of the entire settlement system can grow through agricultural intensification. The first development is entirely concerned with the organization of the settlement system,
whereas the second development is concerned with the correlation between agricultural intensification and population growth within the entire settlement system. This means that agricultural
intensification will occur once a certain population threshold is surpassed, irrespective of how this
population is dispersed across the landscape.
101 Although this explanation of radial lines around northern Mesopotamian sites is widely accepted, other hypotheses
exist. It has been suggested that the radial lines functioned as water harvesting networks, although this explanation
does not necessarily rule out the possibility that they functioned as roads at some point in time (McLellan and Porter
1995; McLellan et al. 2000). Weiss doubted the ancient nature of the hollow ways and suggested that they were of
much more recent date (Ristvet 2005: 26; Weiss 2007).
102 Ethnographic research indicates that manure derived from human settlements usually comes from animal pens and
stables where few other activities are carried out, and that where manure is collected from open-air spaces, it is cleaned
before being transported to the fields (Horne 1994: 53). This procedure would also reduce weight. An alternative
explanation of sherd scatters is that they derived from an efficient type of irrigation called pitcher-irrigation, which
was traditionally used in India and Iran (Barrow 1987: 240). Pitcher irrigation would leave an archaeological residue
consisting of sherds spread across the landscape. If this were the case, one would expect the repertoire of pot shapes
to be more limited, i.e. biased toward closed forms, than when these scatters would result from communal site refuse
collected for manuring purposes. Additionally, because pitchers must have water-permeable walls, it is possible to
test whether sherds could have resulted from pitcher irrigation even in the absence of good rims and other feature
sherds.
143
Given that the total amount of settled area remained roughly the same, it is theoretically possible that the Tell al-Hawa region operated as a closed system free of any external influences or population movements (Wilkinson and Tucker 1995: 51). This is also suggested by the present analysis,
which found that the NJS region was characterized by a relatively stable population growth falling
entirely within the expected natural growth rates (see Section 5.3.1). The transition from a Late
Chalcolithic ‘resilient risk-reduction subsistence system’ toward a mid–late third millennium ‘brittle product-maximization system’ (Wilkinson and Tucker 1995: 85) can therefore be fully explained
in terms of natural population growth and associated agricultural intensification (see Boserup
1965).
The formation of a site hierarchy during the mid to late third millennium bc, and its continued
existence during the early second millennium bc remains, however, unexplained. There must have
been clear advantages for either the central site to assert its control over its neighbours, or for secondary sites to relinquish their surplus production to another settlement. Especially during periods
of resource stress, second-rank settlements may have actively resisted the extraction of their agricultural surplus by the central site in the settlement system. In other words, why did a hierarchical
settlement system arise, rather than a system of more independent and equal-sized settlements,
which would be expected under conditions of increased resource stress? Wilkinson recently suggested that the emergence of anomalously large sites could have been stimulated by the formation of inter-regional exchange networks (2000b: 12; 2004: 186). A complex road network spread
across northern Mesopotamia during the third and second millennium bc (Lebeau 2000b; Ur 2003;
Marro 2004b, a), and many sites along these roads attained a size that cannot be explained solely by
the subsistence potential of these sites’ catchment areas. These networks would have been focused
on the exchange of high-value commodities, as well as exchange between sedentary and (semi-)
nomadic population components. The location of certain sites along important roads may have
provided these sites with an advantage in resource access over neighbouring sites. Eventually, this
advantage may have developed into unequal relations between settlements, whereby staples and
high-value commodities moved in different directions, thereby creating a highly flexible economy
combining agricultural, pastoral, and high-value commodities (Wilkinson 2000b: 14–16).
During the early second millennium bc, the number of settlements increased, and average settlement size declined, although Tell al-Hawa retained its status as the largest site in the region.
The 2000 bc rank-size plot is slightly primate (Fig. 5.13C), whereas the 1500 bc plot shows a slight
tendency towards a primo-convex distribution (Fig. 5.13D), although the large error-margin inhibits any further comments being made on these plots. Again, as with the late third millennium
settlement pattern, the observed distributions for the early second millennium bc are not easily
explained by the expected patterns for inter-settlement social relations. One possibility is that Tell
al-Hawa retained the key function in the inter-regional exchange network that it acquired in the
third millennium, and could therefore continue to assert its influence over neighbouring communities, thereby neglecting the effects of stronger inter-site competition. The close proximity of the
NJS to the large sites along the Tigris makes it likely that the area profited from the Old Assyrian
trade network extending into Anatolia (Wilkinson and Tucker 1995: 55).
Above, it was already suggested that large central sites not only developed in response to interregional exchange, but also played a role in exchange between sedentary and (semi-)nomadic communities. Although this topic will be more fully explored in the following section, some preliminary
comments are necessary to explain the continued role of Tell al-Hawa as the dominant site in the
NJS region during the early second millennium bc. It was argued that the extensive urbanization
that characterized large parts of the Upper Khabur, as well as the NJS region, necessitated a relocation of specialized pastoral production toward non-urban areas (Section 6.5.1). This process, in
turn, may have stimulated the emergence of central sites in the exchange between (semi-)nomadic
pastoralist communities and sedentary agriculturalist communities. It may therefore be tentatively
suggested that it was exactly this function as gateway city (see Burghardt 1971) between urbanized areas and more marginal areas frequented by (semi-)nomadic communities that allowed Tell
al-Hawa, and sites like Tell Brak and Mari to continue to exist as important settlements dominating
their environments. Thus, when aridity set in during the late third millennium, these settlements
capitalized on their functions as gateway cities which they had acquired earlier that millennium.
144
Summarizing, natural population growth and the formation of exchange networks were the
determining factors in shaping the settlement patterns that were observed in the NJS region.
The initial formation of long-distance exchange networks during the mid–late third millennium,
propelled by a demand for high-value and/or pastoral commodities, may have led to the prominence of Tell al-Hawa within the larger settlement system. It was probably exactly this central and
perhaps monopolizing position with respect to exchange with (semi-)nomadic communities that
allowed Tell al-Hawa to continue dominating its surroundings when late third–early second millennium bc increased aridity necessitated a stronger focus on agriculturalist-pastoralist interaction.
In this sense, the onset of drier conditions reinforced rather than changed a social constellation
that was already highly geared towards maintaining an urbanized society in a relatively marginal
environment.
8.3 The development of sedentary-(semi-)nomadic relations
It has been argued that groups relying on different subsistence strategies would develop cooperative mechanisms in order to increase each group’s chance of survival and access to contested and
therefore scarce resources (Section 3.5.2). This expectation is based on the concept that specialization is on the one hand more productive than diversification, but on the other hand makes a group
more vulnerable to environmental stress. Chapters 6 and 7 reviewed the evidence for the existence
of a pastoral economy in northern Mesopotamia, as well as the evidence for the existence of a social identity of these pastoralists. The present section will discuss to what degree the reconstructed
developments conform to the expected patterns.
Chapter 6 presented a scenario for the development of the pastoral economy in northern
Mesopotamia during the third and early second millennium bc. It was suggested that a complex
and fully developed pastoral economy had existed in Mesopotamia since the fourth millennium
bc. During the mid-third millennium, urbanization in the Upper Khabur led to a relocation of
pastoral production to more marginal areas to the south and to the west of the main urbanized
areas of the central and eastern Upper Khabur, that is, the Middle Khabur and Jebel Abd al-Aziz
areas, and the area around Tell Beydar, respectively. Eventually, the small-scale pastoral production along the Middle Khabur was supplemented by specialized production focused on the area of
the Jebel Abd al-Aziz. This specialized distant village-based or transhumant pastoral production
system ended somewhere during the late third millennium and was probably replaced by a system
whereby specialized pastoral production was practised by more mobile, (semi-)nomadic communities. Early second millennium bone assemblages from northern Mesopotamia are scarce, but the
textual evidence from Mari testifies to the importance of the pastoral economy to society overall,
and to the high degree of mobility of pastoralist communities.
Chapter 7 discussed the evidence for the presence of Amorites in greater Mesopotamia, and
aimed to identify the nature of Amorite identity, as well as its diachronic development. It was
demonstrated that Amorite can be considered an ethnic identity during the early second millennium, and that this ethnic connotation also existed during earlier periods. This identity was
expressed in a vocabulary focussing on a pastoralist ideal, even though being pastoralist was not
essential to being Amorite. Textual references to Amorites show that they were present in greater
Mesopotamia from at least the middle third millennium bc onwards, and that they operated on all
levels of society. During the Ur III period, a dichotomy existed between public records displaying
Amorites as non-human enemies of the state, and administrative records indicating that Amorites
held all manner of positions in society. During the early second millennium, states emerged across
Mesopotamia, led by rulers who emphasized their Amorite heritage. The evidence from Mari has
shown that these states were based on tribal identities including both agriculturalist and pastoralist
communities.
145
If these developments and the evidence for climate change at the end of the third millennium
are examined more closely, it can be observed that there is a considerable degree of synchronicity between important changes in the pastoral economy, the Amorite identity, and climate change.
With respect to the expected interaction between sedentary and (semi-)nomadic communities, the
following scenario is suggested.
The date of the genesis of the Amorite identity cannot be fixed with certainty on the basis of
the presently available evidence. Textual sources indicate that the Amorite ethnicity was recognized
by the mid-third millennium bc, but it may have been present long before that time. Its increased
importance towards the end of the third millennium may be associated with the trend towards
specialization in the pastoralist economy that occurred during the mid-third millennium bc. This
trend may have stimulated the emergence of an ethnic identity through which exchanges between
different economic groups were carried out. This development was most pronounced in, although
not exclusive to, in the wider Khabur area, making this region a potential candidate for an Amorite
homeland, or part of it. The Amorite identity was adopted by individuals and groups who found
the identity’s potential for pastoralist-agriculturalist interaction advantageous. This advantage may
have derived from the increased sensitivity to environmental stress of both the pastoral as well
as the agricultural economy, which was a direct consequence of the development toward stronger
specialization. However, as long as this economic system functioned properly, the Amorite identity may have either constituted a minority within Mesopotamian society at large, or Amorites may
not have felt the need to emphasize their identity and to invoke ethnic solidarity. The second interpretation is favoured here because it does not necessarily exclude the possibility that Amorites
indeed represented a minority, and because it accords well with the idea that ethnicity is a situational identity that is usually only emphasized under specific conditions. The relative scarcity of
references to Amorites before the Ur III period should therefore not be understood as an absence
of Amorites, but as an indication that the Amorite identity was not emphasized by its carriers, and
that scribes found it unimportant to record this aspect of identity. This interpretation, then, suggests that Amorites may have constituted a significant ethnic group within parts of Mesopotamia
well before the Ur III period.
The window into Mesopotamian history and society that is provided by the Ur III textual records clearly shows an entirely different picture. At that time, the Amorite ethnicity had
certainly become an important aspect of identity, and many Amorites lived throughout greater
Mesopotamia. Rather than arguing that the sudden increase in Amorite names and references during the late third/early second millennium bc resulted from population displacements, it is here
suggested that this increase reflected the flexible and situational nature of ethnicity as identity. As
argued earlier, ethnicity is an identity that can be adapted to very specific needs and situations. The
growing importance of the Amorite identity from the Ur III period onward must therefore be explained in terms of the advantages associated with becoming Amorite.
As argued in Section 7.3.3, the Amorite identity constituted a social network between pastoralist and agriculturalist communities that was primarily defined through an idiom of pastoralist
ideals. The growth of this network is visible in the Ur III records and resulted in the complex
early second millennium bc social landscape of numerous Amorite kingdoms throughout greater
Mesopotamia. It coincided with climate change and increasing aridity during the late third–early
second millennium bc onward. Increasing environmental stress may have strained the agricultural
and pastoral economies, and prompted closer cooperation and contact between the two economies
in order to create a more diverse and reliable resource base. At the same time, the higher mobility of pastoralist communities that is suggested by the early second millennium textual evidence
impeded regular contacts and exchanges, thereby reinforcing the need for a social network that
strengthened ties between agriculturalist and pastoralist communities. It is in this sense, then, that
the growing importance of the Amorite identity should be interpreted. Through this identity,
Amorites could create an extensive exchange network that was organized along constructed or perceived kinship lines and based on ethnic solidarity and more generalized reciprocity than would be
possible among ethnically differentiated groups. The Amorite identity and its associated values and
the social network it embodied represented a behavioural alternative that could be implemented
under certain circumstances (see Section 3.4). Eventually, this adaptive advantage not only allowed
146
Amorite groups to be more successful than their competitors, but would also generate an ethnic
shift toward becoming Amorite, eventually resulting in the patchwork of Amorite kingdoms that
can be observed for the early second millennium bc.
The result of this process of becoming Amorite is most clearly seen in the Mari archives that
allow the most detailed view of the organization of an Amorite state. As was argued in Section
7.2.3, this archive has shown the importance of tribal affiliation, and it has recently been shown
that many conflicts were not fought over the nomadic pastoralist-sedentary agriculturalist or tribestate divide, but over tribal boundaries. This interpretation is in line with the interpretation offered
here, namely that the Amorite identity distinguished itself from other identities primarily through
its ability to combine agriculturalist and pastoralist production modes. The unique view afforded
by the Mari archives possibly allows us to propose that these tribes functioned as factions within
the Amorite ethnicity, in the sense proposed by Brumfiel (1994). It could be tentatively suggested
that such factions emerged out of a desire by individual city-states to attach themselves to particular (semi-)nomadic groups, or vice versa, thereby creating stronger social and reciprocal obligations. The strict association of settlements with groups such as the Sim’alites or the Yaminites
(see Section 5.4.2) suggests that factionalism may have been significant in the Mari kingdom, as
is also suggested by the frequent conflicts between these tribal confederacies (see Section 7.2.3).
Ristvet recently combined textual, geological, and archaeological evidence to locate pasture areas
in the Upper Khabur (2008), and it might be tentatively suggested that distinct areas were closely
associated with individual cities.
Because the Amorite identity is associated with specialized pastoralism, it may also be associated with urbanism. Throughout history, increased demographic stress seems to have been one
of several processes that can induce specialization of pastoral production. The Amorite identity
could be understood as the social reflection of this causal relationship between urbanism and
specialized pastoralism, and the social element that binds these components together (cf. Porter
2004: 73-4). The increased importance of the Amorite identity would, then, not only reflect social
interaction between agricultural and pastoral communities in general, but it would also be a reflection of an urban culture that tried to maintain its identity and existence in the face of climate
change. Additionally, the Amorite identity may eventually have eased transitions between these
components by bridging social and ethnic boundaries and differences in social status. The identity
thus accommodated the demographic fluidity that can be observed in more recent Near Eastern
societies, as described, for example, for the Biqa Valley (Marfoe 1979). The spread of the Amorite
identity may therefore have not only signified a reorganization of economic relations between different social units, but possibly also a re-evaluation of each unit’s social status.
Summarizing, it has been argued that the Amorite ethnic identity represented a social network between pastoralist and agriculturalist communities that emerged as a response to increasing
aridity and associated environmental stress. The explicitly kinship-based ethnic identity allowed
Amorite groups to participate in an exchange network based on ethnic solidarity and generalized reciprocity. The success of these groups eventually resulted in an ethnic shift whereby other
groups sought to associate themselves with this network by becoming Amorite. The emergence of
Amorite kingdoms and dispersal of Amorite names across greater Mesopotamia should therefore
not be understood as evidence for moving people, but as evidence for a moving identity.103
103 The Amorite ethnicity was not the only ethnic identity to appear in greater Mesopotamia at the end of the third millennium. The records from Ebla do not include Hurrian names, and these seem to be absent in the Beydar archives as
well. However, certainly from the time of Naram-Sin, Hurrians were present in the Upper Khabur area. Whereas the
Amorite identity was more strongly focused on northern Mesopotamia and the southern alluvium, the Hurrian identity
probably extended northwards into the Caucasus (Steinkeller 1998). It is interesting to note that the boundary between
the Hurrian and Amorite identity must have been located in northern Mesopotamia, an area that was extremely sensitive to the climatic fluctuations that led to the increasing importance of the Amorite identity, and that both identities
extended into distinct environmental zones. It could very tentatively be suggested that this ethnic boundary may have
shifted north and south as a result of groups aligning themselves with either Hurrian or Amorite neighbours, depending on the adaptive advantage inherent in each alternative. This situation would reflect the nature of ethnic change
described by Barth for the border region of Afghanistan and Pakistan (1969b).
147
8.4 Conclusions
The previous analysis of diachronic settlement developments in three regions of northern
Mesopotamia has resulted in three different scenarios of changing human-environment interaction
during the late third–early second millennium bc. In each region, natural population growth and
the effects of mid-third millennium urbanization were found to be much more important factors
in determining human-environment interaction than increasing aridity during the late third–early
second millennium bc. Where social responses to climate change could be detected, as was the case
for the B-EDS and the NJS, these responses were found to be primarily reinforcing trends that
had already been set in motion as a result of mid-third millennium urbanization. Furthermore, the
expected pattern of inter-site competition, driven by resource stress, was only encountered in the
BS region, and there it was primarily correlated to population growth, with increasing aridity only
acting as a reinforcing mechanism. In the NJS region, the prior existence of an exchange network
focused on Tell al-Hawa probably ensured the continuing dominance of that site and prevented
the emergence of inter-site competition.
The analysis of pastoralist-agriculturalist interaction has shown that the development of the
Amorite identity can be explained by the theoretical framework presented in Section 3.5. The
Amorite identity functioned as a social network that enabled pastoralist and agriculturalist groups
to maintain stable exchange relations. These relations provided Amorite groups with mutual access to diverse resources, thereby stabilizing the groups’ respective economies. In this sense, the
adoption of the Amorite identity throughout the Near East during the late third–early second millennium may have been fundamental to the creation of the resilient dimorphic economy that was
the hallmark of the Mari state.
Concluding, the present study has shown that climate change impacted upon human societies
in marginal areas of the Near East, but also that careful analysis of local environmental, cultural,
and economic conditions is required in order to distinguish climate-induced changes from other
developments. It has been shown that a good understanding of prevailing socio-cultural conditions is required in order to understand why a human group does, or does not respond to climate
change as expected. More importantly, this study has made clear that local climate change scenarios, as for example presented in the Leilan Climate Change Model (see Section 1.3), can rarely if
ever be extrapolated to other regions or periods without major adjustments. This observation also
suggests that although lessons may be drawn from the past, the past does not dictate the future.
Finally, the case-studies analyzed here, and the diversity of identified responses, show above all the
extreme resiliency of human communities when faced with climate change.
8.5 Final comments: looking back and looking into the future
This study has shown that effective research into responses to environmental change requires (a)
a good chronological control of both palaeoclimatic as well as culture-historical data, (b) a clear
understanding of the effects of environmental change on human communities, (c) proof that past
human groups would disappear if they failed to adapt to environmental change, and (d) a theoretical framework that helps to model expected responses to environmental change in order to
detect environmentally triggered socio-cultural developments in the multitude of developments
that make up cultural change (see Chapters 1 and 3). By evaluating the degree to which the present
study has succeeded in fulfilling these requirements, it is possible to highlight areas of potential further research to increase understanding of human-environment interaction in northern
Mesopotamia during the late third–early second millennium bc.
Chronological control of both palaeoclimate proxy records as well as archaeological and historical chronologies remains problematic. The lack of local well-dated palaeoclimate proxy records
is unlikely to be resolved in the near future. A relatively high precision is now being achieved for
the Soreq Cave and Lake Van proxy records. Nevertheless, the Soreq Cave records rely on Useries dates, which were taken at intervals and from which an average sedimentation rate was calculated for the intervals. This means that dating unique climate events rather than long-term environmental trends remains problematic and associated with a considerable degree of uncertainty.
148
Archaeological and cultural chronologies in northern Mesopotamia generally suffer from a lack of
high-quality radiocarbon dates. The continuing debates on the Early Jezirah (EJ) periodization and
the earliest date for Khabur ware, the definition of characteristic pottery shapes for each EJ subperiod and each northern Mesopotamian culture area, and the correlation between material culture
and historical chronology demonstrate that an independent high-precision radiocarbon chronology for Mesopotamia is becoming increasingly vital. Until the establishment of such a chronology,
the drawing of correlations between unique climate events and unique historical events should be
avoided, and it is perhaps better to try to reconcile long-term processes, for which both the palaeoclimate and archaeological data are currently better suited.
The reconstruction of the actual effects of environmental change is impeded by the lack highprecision palaeoclimate proxy records for northern Mesopotamia. The local northern Mesopotamian
records all point to a significant shift towards aridity somewhere during the third millennium bc
(see Section 2.3.5). However, the available records usually only indicate a relative shift and cannot be used to calculate changes in annual rainfall or river discharges. Furthermore, the detected
shifts were only precisely dated by comparison with more precise, but more distant proxy records,
which may have been unaffected by some of the more regional northern Mesopotamian climatic
fluctuations. Although recent climate modelling using better dated but more distant proxy records
is starting to fill this gap, the quality and resolution of these models is still relatively low and their
validity remains to be tested. This lack of high-precision palaeoclimate data reinforces the need to
test whether climate change was actually sufficiently severe to threaten human life-ways. This study
has collected some evidence that this might have been the case (Section 2.3.8), but there is a need
for site-specific as well as regional studies dealing with differences across northern Mesopotamia.
Finally, there is a potentially large diversity in responses to environmental change. For this reason, it is necessary to consider the full range of possible responses, instead of focusing on a single
response and present an argument based on its occurrence or absence. As this study has shown,
reliance on settlement data alone provides a distorted view of northern Mesopotamian humanenvironment interaction. The reason for this distortion must be sought in the fact that at least a
part of the total response structure took place beyond the realm of settled life, namely in those
parts of northern Mesopotamia that were home to (semi-)nomadic communities. This complex
interaction, and the changes therein, has only been detected through the adoption of a theoretical
framework that explicitly aimed to address a wide range of responses, and predict which response
is adopted under changing environmental conditions. In this sense, then, past climate changes
perhaps challenge today’s archaeologists as much as that these changes challenged the peoples of
the past.
149
Bibliography
Abdi, K. 2003: The early development of pastoralism in the Central Zagros Mountains. Journal of World
Prehistory 17/4, 395–448.
Abruzzi, W.S. 1982: Ecological theory and ethnic differentiation among human populations. Current
Anthropology 23/1, 13–35.
Abu Assaf, A. 1978/9: Archäologische Geländebegehung in den Bezirken Homs, Tartus, Lataqiya und
Raqqa. Archiv für Orientforschung 26, 176–7.
Adams, R.M. 1965: Land behind Baghdad: a history of settlements on the Diyala Plains. Chicago.
—. 2006: Shepherds at Umma in the Third Dynasty of Ur: interlocutors with a world beyond the
scribal field of ordered vision. Journal of the Economic and Social History of the Orient 49/2, 133–69.
Akkermans, P.M.M.G. 1984: Archäologische Geländebegehung im Balih-Tal. Archiv für Orientforschung
31, 188–90.
—. 1993: Villages in the steppe. Later Neolithic settlement and subsistence in the Balikh valley, northern Syria. Ann
Arbor.
Akkermans, P.M.M.G. and G.M. Schwartz 2003: The archaeology of Syria. From complex hunter-gatherers to
early urban societies (ca. 16,000–3000 BC). Cambridge World Archaeology. Cambridge.
Algaze, G., R. Breuninger, and J. Knudstad 1994: The Tigris-Euphrates archaeological reconnaissance
project: final report on the Birecik and Carchemish Dam survey areas. Anatolica 20, 1–96.
Algaze, G., R. Breuninger, C. Lightfoot, and M. Rosenberg 1991: the Tigris-Euphrates archaeological
reconnaissance project: a preliminary report of the 1989–1990 seasons. Anatolica 17, 175–240.
Altaweel, M. 2006: Excavations in Iraq: the Ray Jazirah Project, first report. Iraq 68, 155–81.
—. 2007: Excavations in Iraq: the Jazirah Salvage Project, second report. Iraq 69, 117–44.
—. 2008: Investigating agricultural sustainability and strategies in northern Mesopotamia: results
produced using a socio-ecological modeling approach. Journal of Archaeological Science 35/4, 821–35.
Ammerman, A.J. 1981: Surveys and archaeological research. Annual Review of Anthropology 10, 63–88.
Anastasio, S., M. Lebeau, and M. Sauvage 2004: Atlas of preclassical Upper Mesopotamia. Subartu, 13.
Turnhout.
Anbar, M. 1991: Les tribus amurrites de Mari. Orbis Biblicus et Orientalis, 108. Freiburg.
Archi, A. 1985: Mardu in the Ebla texts. Orientalia 54, 7–13.
Archi, A. and M.G. Biga 2003: A victory over Mari and the fall of Ebla. Journal of Cuneiform Studies 55,
1–44.
Arz, H.W., F. Lamy, and J. Patzold 2006: A pronounced dry event recorded around 4.2 ka in brine
sediments from the Northern Red Sea. Quaternary Research 66/3, 432–41.
Aurenche, O. 1981: Essai de démographie archéologique. L’exemple des villages du Proche-Orient
ancien. Paléorient 7/1, 93–105.
Aurenche, O. and P. Desfarges 1983: Travaux d’ethnoarchéologie en Syrie et Jordanie. Rapports
préliminaires. Syria 60, 147–85.
Bahrani, Z. 2006: Race and ethnicity in Mesopotamian antiquity. World Archaeology 38/1, 48–59.
Baillie, M.G.L. 1998: Review of Dalfes, H. N., G. Kukla and H. Weiss: third millennium bc climate
change and Old World collapse. Journal of Archaeological Science 25/2, 185–6.
Ball, W., D.J. Tucker, and T.J. Wilkinson 1989: The Tell al-Hawa project. Archaeological investigations
in the north Jazira 1986–87. Iraq 51, 1–66.
151
Balossi, F., G.M. Di Nocera, and M. Frangipane 2007: The contribution of a small site to the study of
settlement changes on the Turkish Middle Euphrates between the third and second millennium
B.C: preliminary stratigraphic data from Zeytinli Bahçe Höyük. In C. Kuzucuoğlu and C. Marro
(eds.): Sociétés humaines et changement climatique à la fin du troisième millénaire: une crise a-t-elle eu lieu en
Haute Mésopotamie? Varia Anatolica, 19. Istanbul, 355–81.
Banning, E.B. 2002: Archaeological survey. Manuals in Archaeological Method, Theory, and Technique.
New York.
Bar-Matthews, M., A. Ayalon, M. Gilmour, A. Matthews, and C.J. Hawkesworth 2003: Sea-land oxygen
isotopic relationships from planktonic foraminifera and speleothems in the Eastern Mediterranean
region and their implication for paleorainfall during interglacial intervals. Geochimica et Cosmochimica
Acta 67/17, 3181–99.
Bar-Matthews, M., A. Ayalon, and A. Kaufman 1997: Late Quaternary paleoclimate in the Eastern
Mediterranean region from stable isotope analysis of speleothems at Soreq Cave, Israel. Quaternary
Research 47/2, 155–68.
Bar-Matthews, M., A. Ayalon, and A. Kaufman 1998: Middle to Late Holocene (6,500 yr. period)
paleoclimate in the Eastern Mediterranean region from stable isotopic composition of
speleothems from Soreq Cave, Israel. In A.S. Issar and N. Brown (eds.): Water, environment and society
in times of climatic change. Water Science and Technology Library, 31. Dordrecht, 203–14.
Barfield, T.J. 1993: The nomadic alternative. Englewood Cliffs.
Barge, O. and B. Moulin 2008: The development of the Syrian steppe during the Early Bronze Age. In
H. Kühne, R.M. Czichon, and F.J. Kreppner (eds.): Proceedings of the 4th International Congress of the
Archaeology of the Ancient Near East, 1. Wiesbaden, 19–28.
Barnard, A. 1992: Social and spatial boundary maintenance among Southern African hunter-gatherers.
In M.J. Casimir and A. Rao (eds.): Mobility and territoriality. Social and spatial boundaries among foragers,
fishers, pastoralists and peripatetics. New York, 137–51.
Barrow, C. 1987: Water resources and agricultural development in the tropics. Longman Development Studies.
Burnt Mill.
Barth, F. 1969a: Introduction. In F. Barth (ed.): Ethnic groups and boundaries. The social organization of
culture difference. Boston, 9–38.
—. 1969b: Pathan identity and its maintenance. In F. Barth (ed.): Ethnic groups and boundaries. The social
organization of culture difference. Boston, 117–34.
Bartl, K. 1994: Frühislamische Besiedlung im Balih-Tal, Nordsyrien. Berliner Beiträge zum Vorderen Orient,
15. Berlin.
Bates, D.G. and S.H. Lees 1977: The role of exchange in productive specialization. American
Anthropologist 79/4, 824–41.
Battini-Villard, L. 1999: L’espace domestique en Mésopotamie de la IIIe dynastie de’Ur à l’époque paléobabylonienne. BAR International Series, 767. Oxford.
Bauer, J., R.K. Englund, and M. Krebernik 1998: Mesopotamien. Späturuk-Zeit und Frühdynastische Zeit.
Orbis Biblicus et Orientalis, 160/1. Freiburg.
Beaulieu, P.-A. 2005: The god Amurru as emblem of ethnic and cultural identity. In W.H. van Soldt,
R. Kalvelagen, and D. Katz (eds.): Ethnicity in ancient Mesopotamia: papers read at the 48th Rencontre
Assyriologique Internationale, Leiden, 1–4 July 2002. Uitgaven van het Nederlands Instituut voor het
Nabĳe Oosten te Leiden, 102. Leiden, 31–46.
Bell, B. 1971: The dark ages in ancient history. I. The first dark age in Egypt. American Journal of
Archaeology 75/1, 1–26.
Bentley, G.C. 1987: Ethnicity and practice. Comparative Studies in Society and History 29/1, 24–55.
Bernbeck, R. 1993: Steppe als Kulturlandschaft: das ‘Agig-Gebiet Ostsyriens vom Neolithikum bis zur islamischen
Zeit. Berliner Beiträge zum Vorderen Orient. Ausgrabungen, 1. Berlin.
152
Bibliography
Besonen, M. and M. Cremaschi n.d.: Geomorphological field survey report. Tell Leilan, June 2002.
Available at: http://leilan.yale.edu/works/geo_report/index.html (Accessed 10 April, 2008).
Bilgen, A.N. 2001: Early Bronze Age vessels (grave goods?) from Harabebezikan Höyük. In N. Tuna, J.
Öztürk, and J. Velibeyoğlu (eds.): Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish
Dam Reservoirs. Activities in 1999. Ankara, 437–52.
Binford, L.R. 2001: Constructing frames of reference. An analytical method for archaeological theory building using
hunter-gatherer and environmental data sets. Berkeley.
Bintliff, J.L. 1997: Further considerations on the population of ancient Boeotia. In J.L. Bintliff (ed.):
Recent developments in the history and archaeology of central Greece: proceedings of the 6th International
Boeotian conference. BAR International Series, 666. Oxford, 231–52.
—. 2000: The concepts of ‘site’ and ‘offsite’ archaeology in surface artefact survey. In M. Pasquinucci
and F. Trément (eds.): Non-destructive techniques applied to landscape archaeology. The Archaeology of
Mediterranean Landscapes, 4. Oxford, 200–15.
Bintliff, J.L., M. Kuna, and N. Venclová (eds.) 2000: The future of surface artefact survey in Europe. Sheffield
Archaeological Monographs, 13. Sheffield.
Black, J.A., G. Cunningham, J. Ebeling, E. Flückiger-Hawker, E. Robson, J. Taylor, and G. Zólyomi
1998–2006: The Electronic Text Corpus of Sumerian Literature. Available at: http://etcsl.orinst.
ox.ac.uk/ (Accessed 19 November, 2008).
Boerma, J.A.K. 1988: Soils and environment of Tell Hammam et-Turkman. In M.N. van Loon (ed.):
Hammam et-Turkman I. Report on the University of Amsterdam’s 1981–1984 excavations in Syria. Uitgaven
van het Nederlands Historisch-Archaeologisch Instituut te Istanbul, 63. Istanbul, 1–11.
Boessneck, J. 1988: Tierknochenfunde vom Tell Chuēra / Nordost-Syrien. In U. Moortgat-Correns: Tell
Chuēra in Nordost-Syrien. Vorläufige Berichte über die neunte und zehnte Grabungskampagne 1982 und 1983.
Tell Chuēra in Nordost-Syrien, 9. Berlin, 79–98.
Boone, J.L. 1992: Competition, conflict, and the development of social hierarchies. In E.A. Smith and
B. Winterhalder (eds.): Evolutionary ecology and human behavior. Foundations of Human Behavior.
New York, 301–37.
Boserup, E. 1965: The conditions of agricultural growth. The economics of agrarian change under population
pressure. London.
Bottema, S. 1997: Third millennium climate in the Near East based upon pollen evidence. In H.N.
Dalfes, G. Kukla, and H. Weiss (eds.): Third millennium BC climate change and Old World collapse.
NATO ASI Series, I/49. Berlin, 489–515.
Braidwood, R.J. 1937: Mounds in the plain of Antioch: an archeological survey. Oriental Institute Publications,
48. Chicago.
Braun, D.P. and S. Plog 1982: Evolution of “tribal” social networks: theory and prehistoric North
American evidence. American Antiquity 47/3, 504–25.
Brinkman, J.A. 1977 (1964): Appendix: Mesopotamian chronology of the historical period. In A.L.
Oppenheim: Ancient Mesopotamia. Portrait of a dead civilization. Chicago, 335–48.
Brooks, N. 2006: Cultural responses to aridity in the Middle Holocene and increased social complexity.
Quaternary International 151/1, 29–49.
Bruins, H.J. 2001: Near East: towards an integrated 14C time foundation. Radiocarbon 43/3, 1147–54.
Bruins, H.J., J. van der Plicht, and A. Mazar 2003a: 14C Dates from Tel Rehov: Iron-Age chronology,
pharaohs, and Hebrew kings. Science 300/5617, 315–18.
Bruins, H.J., J.J. Akong’a, M.M.E.M. Rutten, and G.M. Kressel 2003b: Drought planning and rainwater
harvesting for arid-zone pastoralists: the Turkana and Maasai (Kenya) and the Negev Bedouin (Israel). NIRP
Research for Policy Series, 17. Amsterdam.
Brumfiel, E.M. 1994: Factional competition and political development in the New World: an
introduction. In E.M. Brumfiel and J.W. Fox (eds.): Factional competition and political development in the
New World. New Directions in Archaeology. Cambridge, 3–13.
153
Bryson, R.A. and R.U. Bryson 1997: High resolution simulations of regional Holocene climate: North
Africa and the Near East. In H.N. Dalfes, G. Kukla, and H. Weiss (eds.): Third millennium BC climate
change and Old World collapse. NATO ASI Series, I/49. Berlin, 565–93.
Bucak, E., R. Dittmann, S.K. Huh, F. Jarecki, S. Kiltz, U. Röttger, and I. Vorontsov 2004: Short report
on the excavations at Şavi Höyük in 2001. In N. Tuna, J. Öztürk, and J. Velibeyoğlu (eds.): Salvage
Project of the Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs. Activities in 2001. Ankara,
174–8.
Buccellati, G. 1966: The Amorites of the Ur III period. Pubblicazioni del Seminario di Semitistica,
Ricerche, 1. Naples.
—. 1990: ‘River bank’, ‘high country’, and ‘pasture land’: the growth of nomadism on the Middle
Euphrates and the Khabur. In S.M. Eichler, M. Wäfler, and D. Warburton (eds.): Tall al-Hamidiyah
2. Orbis Biblicus et Orientalis, 6. Freiburg, 87–117.
—. 1992: Ebla and the Amorites. In C.H. Gordon and G.A. Rendsburg (eds.): Eblaitica: essays on the
Ebla archives and Eblaite language. Volume 3. Publications of the Center for Ebla Research at New
York University. Winona Lake, 83–104.
—. 1997: Amorites. In E.M. Meyers (ed.): The Oxford encyclopedia of archaeology in the Near East. Volume 1.
New York, 107–11.
—. 2008: The origin of the tribe and of ‘industrial’ agropastoralism in Syro-Mesopotamia. In H.
Barnard and W. Wendrich (eds.): The archaeology of mobility. Old World and New World nomadism.
Cotsen Advanced Seminar Series, 4. Los Angeles, 141–59.
Buccellati, G. and M. Kelly-Buccellati 1999: Das archäologische Projekt Tall Mozan/Urkeš. Mitteilungen
der Deutschen Orient-Gesellschaft zu Berlin 131, 7–16.
—. 2000: The royal palace of Urkesh. Report on the 12th season at Tell Mozan/Urkesh: excavations in
area AA, June–October 1999. Mitteilungen der Deutschen Orient-Gesellschaft zu Berlin 132, 133–83.
Buitenhuis, H. 1988: Archeozoölogisch onderzoek langs de Midden-Eufraat. Onderzoek van het faunamateriaal
uit zes nederzettingen in Zuidoost-Turkije en Noord-Syrië daterend van ca. 10.000 BP tot 1400 AD.
Unpublished dissertation, Rijksuniversiteit Groningen.
—. forthcoming: Preliminary report on faunal remains from Tell Hammam et-Turkman, North Syria.
In D.J.W. Meijer (ed.): Hammam et-Turkman II.
Bunnens, G. 2007: Site hierarchy in the Tishrin Dam area and third millennium geopolitics in Northern
Syria. In E. Peltenburg (ed.): Euphrates river valley settlement: the Carchemish sector in the third millennium
BC. Levant Supplementary Series, 5. Oxford, 43–54.
Burghardt, A.F. 1971: A hypothesis about gateway cities. Annals of the Association of American Geographers
61/2, 269–85.
Buringh, P. 1960: Soils and soil conditions in Iraq. Baghdad.
Butterlin, P. 2007: Mari, les šakkanakku et la crise de la fin du troisième millénaire. In C. Kuzucuoğlu
and C. Marro (eds.): Sociétés humaines et changement climatique à la fin du troisième millénaire: une crise a-telle eu lieu en Haute Mésopotamie? Varia Anatolica, 19. Istanbul, 227–45.
Butzer, K.W. 1982: Archaeology as human ecology: method and theory for a contextual approach. Cambridge.
—. 1997: Sociopolitical discontinuity in the Near East C. 2200 B.C.E.: scenarios from Palestine and
Egypt. In H.N. Dalfes, G. Kukla, and H. Weiss (eds.): Third millennium BC climate change and Old
World collapse. NATO ASI Series, I/49. Berlin, 245–96.
Carneiro, R.L. 1970: A theory of the origin of the state. Science 169/3947, 733–38.
Carneiro, R.L. and D.F. Hilse 1966: On determining the probable rate of population growth during the
Neolithic. American Anthropologist 68/1, 177–81.
Casana, J. 2007: Structural transformations in settlement systems of the northern Levant. American
Journal of Archaeology 111/2, 195–221.
Cashdan, E. 1987: Trade and its origins on the Botletli River, Botswana. Journal of Anthropological
Research 43/2, 121–38.
154
Bibliography
—. 2001a: Ethnic diversity and its environmental determinants: effects of climate, pathogens, and
habitat diversity. American Anthropologist 103/4, 968–91.
—. 2001b: Ethnocentrism and xenophobia: a cross-cultural study. Current Anthropology 42/5, 760–65.
Casimir, M.J. 1992: The dimensions of territoriality: an introduction. In M.J. Casimir and A. Rao (eds.):
Mobility and territoriality. Social and spatial boundaries among foragers, fishers, pastoralists and peripatetics.
New York, 1–26.
Castel, C. 2007: L’abandon d’Al-Rawda (Syrie) à la fin du troisième millénaire: premières tentatives
d’explication. In C. Kuzucuoğlu and C. Marro (eds.): Sociétés humaines et changement climatique à la
fin du troisième millénaire: une crise a-t-elle eu lieu en Haute Mésopotamie? Varia Anatolica, 19. Istanbul,
159–78.
Castel, C. and E. Peltenburg 2007: Urbanism on the margins: third millennium BC Al-Rawda in the
arid zone of Syria. Antiquity 81/313, 601–16.
Cauvin, J. 1970: Mission 1969 en Djezireh (Syrie). Bulletin de la Société Préhistorique Française 67, 286–7.
Chamberlain, A. 2006: Demography in archaeology. Cambridge Manuals in Archaeology. Cambridge.
Chapman, J. 1999: Archaeological proxy-data for demographic reconstructions: facts, factoids or
fiction? In J. Bintliff and K. Sbonias (eds.): Reconstructing past population trends in Mediterranean Europe
(3000 BC–AD 1800). The Archaeology of Mediterranean Landscapes, 1. Oxford, 65–76.
Charpin, D. 1987: Šubat-Enlil et le pays d’Apum. MARI 5, 129–40.
—. 1993: La Syrie à l’époque de Mari: des invasions amorites à la chute de Mari. In S. Cluzan, E.
Delpont, and J. Mouliérac (eds.): Syrie. Mémoire et civilisation. Paris, 144–9.
—. 2003: La «toponymie en mirroir» dans le Proche-Orient amorrite. Revue d’Assyriologie et d’Archéologie
orientale 97, 3–34.
—. 2004: Nomades et sédentaires dans l’armée de Mari du temps de Yahdun-Lîm. In C. Nicolle (ed.):
Nomades et sédentaires dans le Proche Orient ancien. Amurru, 3. Paris, 83–94.
Charpin, D. and J.-M. Durand 1986: «Fils de Sim’al»: les origines tribales des rois de Mari. Revue
d’Assyriologie 80, 141–83.
Charpin, D., D.O. Edzard, and M. Stol 2004: Mesopotamien. Die altbabylonische Zeit. Annäherungen 4. Orbis
biblicus et orientalis, 160/4. Freiburg.
Charpin, D. and N. Ziegler 2003: Florilegium marianum V. Mari et le proche-orient à l’époque amorrite: essai
d’histoire politique. Mémoires de N.A.B.U., 6. Paris.
Cherry, J.F. 1979: Four problems in Cycladic prehistory. In J.L. Davis and J.F. Cherry (eds.): Papers in
Cycladic prehistory. UCLAMon, 14. Los Angeles, 22–47.
Clason, A.T. and H. Buitenhuis 1978: A preliminary report on the faunal remains of Nahr el Homr,
Hadidi and Ta’as in the Tabqa Dam region in Syria. Journal of Archaeological Science 5/1, 75–83.
—. 1998: Patterns in animal food resources in the Bronze Age in the Orient. In H. Buitenhuis, L.
Bartosiewicz, and A.M. Choyke (eds.): Archaeozoology of the Near East III. Proceedings of the third
international symposium on the archaeozoology of southwestern Asia and adjacent areas. ARC-Publication, 18.
Groningen, 233–42.
Coombes, P. and K. Barber 2005: Environmental determinism in Holocene research: causality or
coincidence? Area 37/3, 303–11.
Cooper, J.S. 1983: The curse of Agade. Johns Hopkins Near Eastern Studies. Baltimore.
Cooper, L. 2006: Early urbanism on the Syrian Euphrates. New York.
Copeland, L. 1979: Observations on the prehistory of the Balikh Valley, Syria, during the 7th to 4th
millennia B.C. Paléorient 5, 251–75.
—. 1982: Prehistoric tells in the lower Balikh valley, Syria: report on the survey of 1978. Annales
Archéologiques Arabes Syriennes 32, 251–71.
Córdoba, J.M. 1988: Prospección en el valle de río Balih (Syria). Informe provisional. Aula Orientalis
6/2, 149–88.
155
Costanza, R., L. Graumlich, W. Steffen, C. Crumley, J. Dearing, K. Hibbard, R. Leemans, C. Redman,
and D. Schimel 2007: Sustainability or collapse: what can we learn from integrating the history of
humans and the rest of nature? AMBIO: a Journal of the Human Environment 36/7, 522–7.
Courty, M.-A. 1994: Le cadre paléogéographique des occupations humaines dans le bassin du HautKhabur (Syrie du nord-est). Premiers resultats. Paléorient 20/1, 21–59.
—. 2001: Evidence at Tell Brak for the Late EDIII/Early Akkadian air blast event (4 kyr BP). In D.
Oates, J. Oates, and H. McDonald (eds.): Excavations at Tell Brak, 2. Nagar in the third millennium BC.
Cambridge, 367–72.
Courty, M.-A. and H. Weiss 1997: The scenario of environmental degradation in the Tell Leilan region,
NE Syria, during the late third millennium abrupt climate change. In H.N. Dalfes, G. Kukla, and
H. Weiss (eds.): Third millennium BC climate change and Old World collapse. NATO ASI Series, I/49.
Berlin, 107–47.
Cowgill, G.L. 1975: On causes and consequences of ancient and modern population changes. American
Cribb, R. 1991: Nomads in archaeology. New Studies in Archaeology. Cambridge.
Crumley, C.L. 1994: Historical ecology. A multidimensional ecological orientation. In C.L. Crumley
(ed.): Historical ecology. Cultural knowledge and changing landscapes. School of American Research
Advanced Seminar Series. Santa Fe, 1–16.
Cullen, H.M., P.B. deMenocal, S. Hemming, G. Hemming, F.H. Brown, T. Guilderson, and F. Sirocko
2000: Climate change and the collapse of the Akkadian empire: evidence from the deep sea.
Geology 28/4, 379–82.
Curvers, H.H. 1991: Bronze Age society in the Balikh Drainage (Syria). Unpublished dissertation,
Universiteit van Amsterdam.
d’Hont, O. 2004: Entre sédentarité et nomadisme: éléments pour une définition de ces deux termes
pris dans l’histoire du peuplement de la moyenne vallée de l’Euphrate depuis l’avènement de
l’Islam. In C. Nicolle (ed.): Nomades et sédentaires dans le Proche Orient ancien. Amurru, 3. Paris, 13–24.
Danti, M.D. 1997: Regional surveys and excavations. In R.L. Zettler: Subsistence and settlement in a
marginal environment: Tell es-Sweyhat, 1989–1995 preliminary report. MASCA Research Papers in Science
and Archaeology, 14. Philadelphia, 85–94.
—. 2006–7: The expansion of settlement into the Balikh-Euphrates uplands. Available at: http://www.
jezireh.org/sweyhat_regionalsurvey.html (Accessed 7 January, 2008).
de Roche, C.D. 1983: Population estimates from settlement area and number of residences. Journal of
Field Archaeology 10/2, 187–92.
Deckers, K. and S. Riehl 2007: Fluvial environmental contexts for archaeological sites in the Upper
Khabur basin (northeastern Syria). Quaternary Research 67/3, 337–48.
deMenocal, P.B. 2001: Cultural responses to climate change during the late Holocene. Science 292,
667–73.
Deveci, A. and Y. Mergen 1999: Zeytin Bahçeli Höyük 1998: preliminary report. In N. Tuna and J.
Öztürk (eds.): Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs.
Activities in 1998 Ankara, 113–18.
Dewar, R.E. 1991: Incorporating variation in occupation span into settlement-pattern analysis.
American Antiquity 56/4, 604–20.
—. 1994: Contending with contemporaneity: a reply to Kintigh. American Antiquity 59/1, 149–52.
Diamond, J. 2005: Collapse: How societies choose to fail or survive. London.
Dittmann, R., C. Grewe, S.K. Huh, and C. Schmidt 2001: Bericht über einen Survey des Şavi Höyük
im Gebiet des Kargemiş-Stausees, Südost Turkei 1999. In R.M. Boehmer and J. Maran (eds.): Lux
orientis: Archäologie zwischen Asien und Europa. Festschrift für Harald Hauptmann zum 65. Geburtstag.
Internationale Archäologie Studia honoraria, 12. Rahden, 97–111.
156
Bibliography
Dittmann, R., S.K. Huh, T. Mitschang, E. Müller, U. Röttger, C. Schmidt, and D. Wicke 2002: report on
the first campaign of excavations at Şavi Höyük in 2000. In N. Tuna, J. Öztürk, and J. Velibeyoğlu
(eds.): Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs. Activities in
2000. Ankara, 233–40.
Dohmann-Pfälzner, H. and P. Pfälzner 2001: Ausgrabungen der Deutschen Orient-Gesellschaft in
der zentralen Oberstadt van Tall Mozan/Urkeš. Bericht über die in Kooperation mit dem IIMAS
durchgeführte Kampagne 2000. Mitteilungen der Deutschen Orient-Gesellschaft zu Berlin 133, 97–139.
Dossin, G. 1974: Le site de Tuttul-sur-Balîh. Revue d’Assyriologie 68, 25–34.
Drennan, R.D. and C.E. Peterson 2004: Comparing archaeological settlement systems with rank-size
graphs: a measure of shape and statistical confidence. Journal of Archaeological Science 31/5, 533–49.
Durand, J.-M. 1985: La situation historique des šakkanakku: Nouvelle approche. MARI 4, 147–72.
—. 1992: Unité et diversités au Proche-Orient à l’époque amorrite. In D. Charpin and F. Joannès (eds.):
La circulation des biens, des personnes et des idées dan le Proche-Orient ancien. Actes de la XXXVIIIe Rencontre
Assyriologique Internationale (Paris, 8–10 juillet 1991). Paris, 97–128.
—. 1997: Les documents épistolaires du palais de Mari. Tome I. Littératures Anciennes du Proche-Orient, 16.
Paris.
—. 1998: Les documents épistolaires du palais de Mari. Tome II. Littératures Anciennes du Proche-Orient,
17. Paris.
—. 2004a: Mari und die Assyrer. In J.-W. Meyer and W. Sommerfeld (eds.): 2000 v. Chr.: politische,
wirtschaftliche und kulturelle Entwicklung im Zeichen einer Jahrtausendwende. Colloquien der Deutschen
Orient-Gesellschaft, 3. Saarbrücken, 371–92.
—. 2004b: Peuplement et sociétés à l’époque amorrite. (I) Les clans bensim’alites. In C. Nicolle (ed.):
Dyson-Hudson, R. and N. Dyson-Hudson 1980: Nomadic pastoralism. Annual Review of Anthropology 9,
15–61.
Dyson-Hudson, R. and E.A. Smith 1978: Human territoriality: an ecological reassessment. American
Eastwood, W.J., N. Roberts, H.F. Lamb, and J.C. Tibby 1999: Holocene environmental change in
southwest Turkey: a palaeoecological record of lake and catchment-related changes. Quaternary
Science Reviews 18/4–5, 671–95.
Échallier, J.-C. and F. Braemer 1995: Nature et fonctions des «desert kites»: données et hypothèses
nouvelles. Paléorient 21/1, 35–63.
Edzard, D.O. 1997: Gudea and his dynasty. The Royal Inscriptions of Mesopotamia, Early Periods, 3/1.
Toronto.
Eichler, S.M., V. Haas, D. Steudler, M. Wäfler, and D. Warburton 1985: Tall al-Hamidiya 1. Vorbericht
1984. Orbis Biblicus et Orientalis, 4. Freiburg.
Eidem, J. 2000: Northern Jezira in the 18th century BC. Aspects of geo-political patterns. In O.
Rouault and M. Wäfler (eds.): La Djéziré et l’Euphrate syriens de la protohistoire à la fin du IIe millénaire
av. J.-C.: tendances dans l’interprétation historique des données nouvelles. Subartu, 7. Turnhout, 255–64.
Eidem, J., I. Finkel, and M. Bonechi 2001: The third-millennium inscriptions. In D. Oates, J. Oates, and
H. McDonald (eds.): Excavations at Tell Brak, 2. Nagar in the third millennium BC. Cambridge, 99–120.
Eidem, J. and D. Warburton 1996: In the land of Nagar: survey around Tell Brak. Iraq 58, 51–64.
Einwag, B. 1993: Vorbericht über die archäologische Geländebegehung in der Westgazira. Damaszener
Mitteilungen 7, 23–43.
—. 1998: Die Keramik aus dem Bereich des Palastes A in Tall Bi’a/Tuttul und das Problem der frühen Mittleren
Bronzezeit. Münchener Vorderasiatische Studien, 19. München.
Ember, C.R. and M. Ember 1992: Resource unpredictability, mistrust, and war: a cross-cultural study.
Journal of Conflict Resolution 36/2, 242–62.
157
Emberling, G. 1997: Ethnicity in complex societies: archaeological perspectives. Journal of Archaeological
Research 5/4, 295–344.
Enzel, Y., R. Bookman, D. Sharon, H. Gvirtzman, U. Dayan, B. Ziv, and M. Stein 2003: Late Holocene
climates of the Near East deduced from Dead Sea level variations and modern regional winter
rainfall. Quaternary Research 60, 263–73.
Ergenzinger, P.J., W. Frey, H. Kühne, and H. Kurschner 1988: The reconstruction of environment,
irrigation and development of settlement on the Habur in north-east Syria. In J.L. Bintliff, D.A.
Davidson, and E.G. Grant (eds.): Conceptual issues in environmental archaeology. Edinburgh, 108–28.
Evershed, R.P., S. Payne, A.G. Sherratt, M.S. Copley, J. Coolidge, D. Urem-Kotsu, K. Kotsakis, M.
Ozdogan, A.E. Ozdogan, O. Nieuwenhuyse, P.M.M.G. Akkermans, D. Bailey, R.-R. Andeescu, S.
Campbell, S. Farid, I. Hodder, N. Yalman, M. Ozbasaran, E. Bicakci, Y. Garfinkel, T. Levy, and
M.M. Burton 2008: Earliest date for milk use in the Near East and southeastern Europe linked to
cattle herding. Nature, doi: 10.1038/nature07180.
Faivre, X. n.d.: Histoire du peuplement du “Triangle du Habur” au Bronze moyen et au début du Bronze récent:
étude céramologique d’après les fouilles de Tell Mohammed Diyab (1987–1996) et la prospection dans le bassin
occidental du Haut Habur (1989–1997). Unpublished doctoral thesis, Université Charles de GaulleLille.
Falconer, S.E. and S.E. Savage 1995: Heartlands and hinterlands: Alternative trajectories of early
urbanization in Mesopotamia and the southern Levant. American Antiquity 60/1, 37–58.
Falsone, G. and P. Sconzo 2007: The ‘champagne-cup’ period at Carchemish. A review of the
Early Bronze Age levels on the Acropolis Mound and the problem of the Inner Town. In E.
Peltenburg (ed.): Euphrates river valley settlement: the Carchemish sector in the third millennium BC. Levant
Supplementary Series, 5. Oxford, 73–93.
Fielden, K. 1978/9: Archäologische Geländebegehung im Gebiet des Unteren Ğaġğaġ. Archiv für
Orientforschung 26, 172.
Finkelstein, J.J. 1966: The genealogy of the Hammurapi dynasty. Journal of Cuneiform Studies 20, 95–118.
Flannery, K.V. 1976: Sampling by intensive surface collection. In K.V. Flannery (ed.): The early
Mesoamerican village. San Diego, 51–62.
Fleming, D.E. 2004: Democracy’s ancient ancestors: Mari and early collective governance. Cambridge.
Forbes, H. 1989: Of grandfathers and grand theories: the hierarchised ordering of responses to hazard
in a Greek rural community. In P. Halstead and J. O’Shea (eds.): Bad year economics: Cultural responses
to risk and uncertainty. New Directions in Archaeology. Cambridge, 87–97.
Fortin, M. 2000: Économie et société dans la moyenne vallée du Khabour durant la période de
Ninevite 5. In O. Rouault and M. Wäfler (eds.): La Djéziré et l’Euphrate syriens de la protohistoire à
la fin du IIe millénaire av. J.-C.: tendances dans l’interprétation historique des données nouvelles. Subartu, 7.
Turnhout, 111–36.
Frangipane, M., C. Alvaro, F. Balossi, and G. Siracusano 2002: The 2000 campaign at Zeytinlibahçe
Höyük. In N. Tuna, J. Öztürk, and J. Velibeyoğlu (eds.): Salvage Project of the Archaeological Heritage of
the Ilısu and Carchemish Dam Reservoirs. Activities in 2000. Ankara, 83–99.
Frangipane, M., F. Balossi, G.M. Di Nocera, A. Palmieri, and G. Siracusano 2004: The 2001 excavation
campaign at Zeytinlibahçe Höyük: preliminariy results. In N. Tuna, J. Öztürk, and J. Velibeyoğlu
(eds.): Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs. Activities in
2001. Ankara, 35–56.
Frangipane, M. and E. Bucak 2001: Excavations and research at Zeytinlibahçe Höyük, 1999. In N.
Tuna, J. Öztürk, and J. Velibeyoğlu (eds.): Salvage Project of the Archaeological Heritage of the Ilısu and
Carchemish Dam Reservoirs. Activities in 1999. Ankara, 109–131.
Frayne, D. 1990: Old Babylonian period (2003–1595 BC). The Royal Inscriptions of Mesopotamia, Early
Periods, 4. Toronto.
—. 1993a: Sargonic and Gutian periods (2334–2113 BC). The Royal Inscriptions of Mesopotamia, Early
Periods, 2. Toronto.
158
Bibliography
—. 1993b: Ur III period (2112–2004 BC). The Royal Inscriptions of Mesopotamia, Early Periods, 3/2.
Toronto.
—. 2008: Presargonic period (2700–2350 BC). The Royal Inscriptions of Mesopotamia, Early Periods, 1.
Toronto.
Frumkin, A. 2009: Stable isotopes of a subfossil Tamarix tree from the Dead Sea region, Israel, and
their implications for the Intermediate Bronze Age environmental crisis. Quaternary Research 71/3,
319–28.
Frumkin, A., G. Kadan, Y. Enzel, and Y. Eyal 2001: Radiocarbon chronology of the Holocene Dead
Sea: attempting a regional correlation. Radiocarbon 43/3, 1179–89.
Fuensanta, J.G. 2007: The Tilbes Project (Birecik Dam, Turkish Euphrates): the Early Bronze evidence.
In E. Peltenburg (ed.): Euphrates river valley settlement: The Carchemish sector in the third millennium BC.
Levant Supplementary Series, 5. Oxford, 142–51.
Fuensanta, J.G., E. Bucak, E. Crivelli, P. Charvat, and R. Moya 2005: The research of the Tilbes
Project, 2004. Kazı Sonuçları Toplantısı 27.
Fuensanta, J.G., E. Bucak, E. Crivelli, H. Karabulut, H. Sauren, P. Charvat, and R. Moya 2006: The
Tilbes project research in 2005: Surtepe Höyük excavations. Kazı Sonuçları Toplantısı 28, 457–70.
Fuensanta, J.G., P. Charvat, E. Bucak, R. Moya, and M.A. Jimenez 2002a: Tilbes and Tilvez Höyük
salvage excavation report. Kazı Sonuçları Toplantısı 24, 369–76.
Fuensanta, J.G., P. Charvat, E. Bucak, M.A. Jimenez, P. Kvetina, and F. Velinski 2002b: 2001 Surtepe
Höyük salvage excavation report. Kazı Sonuçları Toplantısı 24, 105–12.
Fuensanta, J.G., M.S. Rothman, and E. Bucak 1998: 1997 Salvage excavations at Tilbeş Höyük. Kazı
Sonuçları Toplantısı 20, 207–18.
Fuensanta, J.G., M.S. Rothman, P. Charvat, and E. Bucak 2001: Tilbeş Höyük salvage project
excavation. Kazı Sonuçları Toplantısı 23, 131–44.
Galaty, J.G. 1982: Being “Maasai”; being “People-of-Cattle”: ethnic shifters in East Africa. American
Ethnologist 9/1, 1–20.
Gallet, Y., M. Le Goff, A. Genevey, J.-C. Margueron, and P. Matthiae 2008: Geomagnetic field intensity
behavior in the Middle East between ~3000 BC and ~1500 BC. Geophysical Research Letters 35/
L02307, doi: 10.1029/2007GL031991.
Galvin, K.F. 1987: Forms of finance and forms of production: the evolution of specialized livestock
production in the ancient Near East. In E.M. Brumfiel and T.K. Earle (eds.): Specialization, exchange,
and complex societies. New Directions in Archaeology. Cambridge, 119–29.
Gasche, H., J.A. Armstrong, S.W. Cole, and V.G. Gurzadyan 1998: Dating the fall of Babylon. A reappraisal
of second-millennium chronology (A joint Ghent–Chicago–Harvard project). Mesopotamian History and
Environment Memoirs, 4. Ghent.
Gelb, I.J. 1980: Computer-aided analysis of Amorite. Assyriological Studies, 21. Chicago.
Goodfriend, G.A. 1999: Terrestrial stable isotope records of Late Quaternary paleoclimates in the
eastern Mediterranean region. Quaternary Science Reviews 18/4–5, 501–13.
Göyünç, N. and W. Hütteroth 1997: Land an der Grenze. Osmanische Verwaltung im heutigen türkisch–syrisch–
irakischen Grenzgebiet im 16. Jahrhundert. Istanbul.
Gragson, T. 1998: Potential versus actual vegetation: human behavior in a landscape medium. In W.
Balée and C.L. Crumley (eds.): Advances in historical ecology. Historical Ecology Series. New York,
213–31.
Green, M.W. 1980: Animal husbandry at Uruk in the archaic period. Journal of Near Eastern Studies
39/1, 1–35.
Gremmen, W.H.E. and S. Bottema 1991: Palynological investigations in the Syrian Gazira. In H. Kühne
(ed.): Die rezente Umwelt von Tall Seh Hamad und Daten zur Umweltrekonstruktion der Assyrischen Stadt
Dur-Katlimmu. Berichte der Ausgrabung Tall Seh Hamad, Dur-Katlimmu, 1. Berlin, 105–16.
159
Gronenborn, D. 2005: Klimaforschung und Archäologie. In D. Gronenborn (ed.): Klimaveränderung und
Kulturwandel in neolithischen Gesellschaften Mitteleuropas, 6700–2200 v. Chr. RGZM Tagungen, 1. Mainz,
1–16.
Guichard, M. 2002: Le Šubartum occidental à l’avènement de Zimrî-Lîm. In D. Charpin and J.-M.
Durand (eds.): Florilegium marianum VI. Recueil d’études à la mémoire d’André Parrot. Mémoires de
N.A.B.U., 7. Paris, 119–68.
Halstead, P. and J. O’Shea 1989a: Introduction: cultural responses to risk and uncertainty. In P.
Halstead and J. O’Shea (eds.): Bad year economics: cultural responses to risk and uncertainty. New
Directions in Archaeology. Cambridge, 1–7.
— (eds.) 1989b: Bad year economics: cultural responses to risk and uncertainty. New Directions in Archaeology.
Cambridge.
Hassan, F.A. 1981: Demographic archaeology. Studies in Archaeology. New York.
Hawkins, J.D. 1980: Karkamis. In D.O. Edzard (ed.): Reallexikon der Assyriologie und Vorderasiatische
Archäologie, V. Berlin, 426–46.
Hempelmann, R. 2008: Kharab Sayyar: the foundation of the Early Bronze Age settlement. In J.M.
Córdoba, M. Molist, M.C. Pérez, I. Rubio, and S. Martínez (eds.): Proceedings of the 5th International
Congress on the Archaeology of the Ancient Near East. Madrid, April 3–8 2006. Vol. II. Madrid, 153–64.
Henry, A.G. and D.R. Piperno 2008: Using plant microfossils from dental calculus to recover human
diet: a case study from Tell al-Raqa’i, Syria. Journal of Archaeological Science 35/7, 1943–50.
Herveux, L. 2004: Étude archéobotanique préliminaire de tell al-Rawda, site de la fin du Bronze ancien
en Syrie intérieure. Akkadica 125/1, 79–91.
Hobbs, J.J. 1990: Bedouin life in the Egyptian wilderness. Cairo.
Hole, F. 1991: Middle Khabur settlement and agriculture in the Ninevite V period. Bulletin of the
Canadian Society for Mesopotamian Studies 21, 17–30.
—. 1997a: Evidence for mid-Holocene environmental change in the Western Khabur Drainage,
Northeastern Syria. In H.N. Dalfes, G. Kukla, and H. Weiss (eds.): Third millennium BC climate change
and Old World collapse. NATO ASI series, I/49. Berlin, 39–66.
—. 1997b: Palaeoenvironment and human society in the Jezireh of northern Mesopotamia 20,000–
6,000 B.P. Paléorient 23/2, 39–49.
—. 1999: Economic implications of possible storage structures at Tell Ziyadeh, NE Syria. Journal of
—. 2009: Pastoral mobility as an adaptation. In J. Szuchman (ed.): Nomads, tribes, and the state in the
ancient Near East. Oriental Institute Seminars, 5. Chicago, 261–83.
Hole, F. and B.F. Zaitchik 2007: Policies, plans, practice, and prospects: Irrigation in northeastern Syria.
Land Degradation & Development 18/2, 133–52.
Horne, L. 1993: Occupational and locational instability in arid land settlement. In C.M. Cameron and
S.A. Tomka (eds.): Abandonment of settlements and regions: ethnoarchaeological and archaeological approaches.
New Directions in Archaeology. Cambridge, 43–53.
—. 1994: Village spaces. Settlement and society in northeastern Iran. Smithsonian Series in Archaeological
Inquiry. Washington.
Horowitz, D.L. 1985: Ethnic groups in conflict. Berkeley.
IJzereef, G.F. 2001: Animal remains. In M.N. Van Loon (ed.): Selenkahiye. Final report on the University
of Chicago and University of Amsterdam excavations in the Tabqa Reservoir, northern Syria, 1967–1975.
Uitgaven van het Nederlands Historisch-Archaeologisch Instituut te Istanbul, 91. Istanbul, 569–85.
IPCC 2007: Climate change 2007: Synthesis report. Contribution of Working Groups I, II and III to the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change. Geneva.
Issar, A.S. and M. Zohar 2004: Climate change. Environment and civilization in the Middle East. Berlin.
160
Bibliography
Jacobsen, T. and R.M. Adams 1958: Salt and silt in ancient Mesopotamian agriculture. Progressive
changes in soil salinity and sedimentation contributed to the breakup of past civilizations. Science
128, 1251–58.
Johnson, A.W. and T.K. Earle 2000: The evolution of human societies. From foraging group to agrarian state.
Second edition. Stanford.
Johnson, G.A. 1977: Aspects of regional analysis in archaeology. Annual Review of Anthropology 6,
479–508.
—. 1980: Rank-size convexity and system integration: a view from archaeology. Economic Geography
56/3, 234–47.
—. 1981: Monitoring complex system integration and boundary phenomena with settlement size data.
In S. Van der Leeuw (ed.): Archaeological approaches to the study of complexity. Cingula, 6. Amsterdam,
144–88.
Jones, S. 1997: The archaeology of ethnicity. Constructing identities in the past and present. London.
Kamp, K.A. and N. Yoffee 1980: Ethnicity in ancient Western Asia during the early second millennium
B.C.: archaeological assessments and ethnoarchaeological prospectives. Bulletin of the American
Schools of Oriental Research 237, 85–104.
Kaptijn, E. 2009: Life on the watershed. Reconstructing subsistence in a steppe region using archaeological survey: a
diachronic perspective on habitation in the Jordan Valley. Dissertation, Leiden University.
Kaptijn, E., L. Petit, E. Grootveld, F. Hourani, G. van der Kooij, and O. al-Ghul 2005: Dayr ‘Alla
regional project: Settling the steppe. First campaign 2004. Annual of the Department of Antiquities of
Jordan 49, 89–99.
Keenan, D.J. 2002: Why early-historical radiocarbon dates downwind from the Mediterranean are too
early. Radiocarbon 44/1, 225–37.
Kelly-Buccellati, M. 2002: Ein hurritischer Gang in die Unterwelt. Mitteilungen der Deutschen OrientGesellschaft zu Berlin 134, 131–48.
Kelly, R.L. 1995: The foraging spectrum. Diversity in hunter-gatherer lifeways. Washington.
Kenyon, K.M. 1966: Amorites and Canaanites. Schweich Lectures on Biblical Archaeology. London.
Khazanov, A.M. 2009: Specific characteristics of Chalcolithic and Bronze Age pastoralism in the Near
East. In J. Szuchman (ed.): Nomads, tribes, and the state in the ancient Near East. Oriental Institute
Seminars, 5. Chicago, 119–27.
Khoury, P.S. and J. Kostiner (eds.) 1990a: Tribes and state formation in the Middle East. Berkeley.
—. 1990b: Introduction: Tribes and the complexities of state formation in the Middle East. In P.S.
Khoury and J. Kostiner (eds.): Tribes and state formation in the Middle East. Berkeley, 1–22.
Kintigh, K.W. 1994: Contending with contemporaneity in settlement pattern studies. American Antiquity
59/1, 143–8.
Klengel, H. 1972: Zwischen Zelt und Palast. Die Begegnung von Nomaden und Seßhaften im alten Vorderasien.
Vienna.
Klengel, H. 1992: Syria: 3000 to 300 B.C.: a handbook of political history. Berlin.
Köhler-Rollefson, I. 1992: A model for the development of nomadic pastoralism on the
Transjordanian Plateau. In O. Bar-Yosef and A.M. Khazanov (eds.): Pastoralism in the Levant.
Archaeological materials in anthropological perspectives. Monographs in World Archaeology, 10. Madison,
11–18.
—. 1996: The one-humped camel in Asia: Origin, utilization and mechanisms of dispersal. In D.R.
Harris (ed.): The origins and spread of agriculture and pastoralism in Eurasia. London, 282–94.
Köhler-Rollefson, I. and G.O. Rollefson 1990: The impact of Neolithic subsistence strategies on the
environment: the case of ‘Ain Ghazal, Jordan. In S. Bottema, G. Entjes-Nieborg, and W. van Zeist
(eds.): Man’s role in the shaping of the Eastern Mediterranean landscape. Rotterdam, 1–14.
161
Kohler, T.A. and C.R. Van West 1996: The calculus of self-interest in the development of cooperation:
sociopolitical development and risk among the Northern Anasazi. In J.A. Tainter and B.B. Tainter
(eds.): Evolving complexity and environmental risk in the prehistoric Southwest. Santa Fe Institute Studies in
the Sciences of Complexity Proceedings, 24. Reading, 169–96.
Kohlmeyer, K. 1981: “Wovon man nicht sprechen kann” Grenzen der Interpretation von bei
Oberflächenbegehungen gewonnenen archäologischen Informationen. Mitteilungen der Deutschen
Orient-Gesellschaft zu Berlin 113, 53–79.
—. 1984: Euphrat-Survey. Die mit Mitteln der Gerda Henkel Stiftung durchgeführte archäologische
Geländebegehung im syrischen Euphrattal. Mitteilungen der Deutschen Orient-Gesellschaft zu Berlin 116,
95–118.
Koliński, R. and J. Piątkowska-Małecka 2008: Animals in the steppe: patterns of animal husbandry
as a reflection of changing environmental conditions in the Khabur Triangle. In H. Kühne, R.M.
Czichon, and F.J. Kreppner (eds.): Proceedings of the 4th International Congress of the Archaeology of the
Ancient Near East, 1. Wiesbaden, 115–27.
Kouchoukos, N.T. 1998: Landscape and social change in late prehistoric Mesopotamia. Dissertation, Yale
University.
Kozbe, G. and M.S. Rothman 2005: Chronology and function at Yarım Höyük, part II. Anatolica
31/111–44.
Kramer, C. 1982: Village ethnoarchaeology: rural Iran in archaeological perspective. Studies in Archaeology.
New York.
Krom, M.D., J.D. Stanley, R.A. Cliff, and J.C. Woodward 2002: Nile River sediment fluctuations over
the past 7000 yr and their key role in sapropel development. Geology 30/1, 71–4.
Kuniholm, P.I., B. Kromer, S.W. Manning, M. Newton, C.E. Latini, and M.J. Bruce 1996: Anatolian tree
rings and the absolute chronology of the eastern Mediterranean, 2220–718 BC. Nature 381/6585,
780–3.
Kupper, J.-R. 1957: Les nomades en Mésopotamie au temps des rois de Mari. Bibliothèque de la Faculté de
Philosophie et Lettres de l’Université de Liège, 142. Paris.
Kuzucuoğlu, C. 2007a: Climatic and environmental trends during the third millennium B.C. in Upper
Mesopotamia. In C. Kuzucuoğlu and C. Marro (eds.): Sociétés humaines et changement climatique à la
459–80.
—. 2007b: Integrating environmental matters in cultural trends. In C. Kuzucuoğlu and C. Marro (eds.):
Sociétés humaines et changement climatique à la fin du troisième millénaire: une crise a-t-elle eu lieu en Haute
Mésopotamie? Varia Anatolica, 19. Istanbul, 21–33.
Landmann, G., A. Reimer, G. Lemcke, and S. Kempe 1996: Dating Late Glacial abrupt climate changes
in 14,570 yr long continuous varve record of Lake Van, Turkey. Palaeogeography, Palaeoclimatology,
Palaeoecology 122, 107–18.
Lawler, A. 2006: North versus South, Mesopotamian style. Science 312, 1458–63.
le Strange, G. 1905: The lands of the eastern Caliphate. Mesopotamia, Persia, and Central Asia from the Moslem
conquest to the time of Timur. Cambridge Geographical Series. Cambridge.
Lebeau, M. 2000a: Stratified archaeological evidence and compared periodizations in the Syrian Jezirah
during the third millennium B.C. In C. Marro and H. Hauptmann (eds.): Chronologies des pays du
Caucase et de l’Euphrate aux IVe–IIIe millénaires. Varia Anatolica, 11. Istanbul, 167–92.
—. 2000b: Les voies de communication en Haute Mésopotamie au IIIe millénaire avant notre ère.
In O. Rouault and M. Wäfler (eds.): La Djéziré et l’Euphrate syriens de la protohistoire à la fin du IIe
millénaire av. J.-C.: tendances dans l’interprétation historique des données nouvelles. Subartu, 7. Turnhout,
157–62.
162
Bibliography
Lemcke, G. and M. Sturm 1997: δ18O and trace element measurements as proxy for the reconstruction
of climate changes at Lake Van (Turkey): preliminary results. In H.N. Dalfes, G. Kukla, and H.
Weiss (eds.): Third millennium BC climate change and Old World collapse. NATO ASI Series, I/49. Berlin,
653–78.
Lévêque, A. n.d.: Atlas de Syrie. Agriculture. Beyrouth.
Lewis, N.N. 1987: Nomads and settlers in Syria and Jordan, 1800–1980. Cambridge Middle East Library.
Cambridge.
—. 1988: The Balikh Valley and its people. In M.N. van Loon (ed.): Hammam et-Turkman I. Report on
the University of Amsterdam’s 1981–84 excavations in Syria. Uitgaven van het Nederlands HistorischArchaeologisch Instituut te Istanbul, 63. Istanbul, 683–695.
Lönnqvist, M. 2006: Archaeological surveys of the Jebel Bishri. The preliminary report of the Finnish
mission to Syria, 2000–2004. Kaskal 3, 203–40.
—. 2008: Were nomadic Amorites on the move? Migration, invasion and gradual infiltration as
mechanisms for cultural transitions. In H. Kühne, R.M. Czichon, and F.J. Kreppner (eds.):
Proceedings of the 4th International Congress of the Archaeology of the Ancient Near East, 2. Wiesbaden,
195–215.
Luke, J.T. 1965: Pastoralism and politics in the Mari period: a re-examination of the character and political
significance of the major West Semitic tribal groups on the Middle Euphrates, ca. 1828–1758 B.C. American
Schools of Oriental Research Dissertation Series, 3. Cambridge (MA).
Lyon, J. 2000: Middle Assyrian expansion and settlement development in the Syrian Jazira: the view
from the Balikh Valley. In R.M. Jas (ed.): Rainfall and agriculture in northern Mesopotamia. Uitgaven van
het Nederlands Historisch-Archaeologisch Instituut te Istanbul, 88. Istanbul, 89–126.
Lyonnet, B. 1996: La prospection archéologique de la partie occidentale du Haut Khabur (Syrie du
nord-est): méthodes, résultats et questions autour de l’occupation aux IIIe et IIe millénaires av. n.
è. Amurru 1, 363–76.
—. 1998: Le peuplement de la Djéziré occidentale au début du 3e millénaire, villes circulaires et
pastoralisme: questions et hypothèses. In M. Lebeau (ed.): About Subartu: Studies devoted to Upper
Mesopotamia. Subartu, 4/1. Turnhout, 179–93.
—. (ed.) 2000: Prospection archéologique du Haut-Khabur occidental (Syrie du N.E.) Volume I. Bibliothèque
Archéologique et Historique, 155. Beyrouth.
—. 2004: Le nomadisme et l’archéologie: problèmes d’identification. Le cas de la partie occidentale
de la Djéziré aux 3ème et début du 2ème millénaire avant notre ère. In C. Nicolle (ed.): Nomades et
sédentaires dans le Proche Orient ancien. Amurru, 3. Paris, 25–49.
Madella, M. and D.Q. Fuller 2006: Palaeoecology and the Harappan civilisation of South Asia: A
reconsideration. Quaternary Science Reviews 25/11–12, 1283–1301.
Magny, M., B. Vanniere, G. Zanchetta, E. Fouache, G. Touchais, L. Petrika, C. Coussot, A.-V. WalterSimonnet, and F. Arnaud 2009: Possible complexity of the climatic event around 4300–3800 cal.
BP in the central and western Mediterranean. The Holocene 19/6, 823–33.
Maisels, C.K. 1990: The emergence of civilization. From hunting and gathering to agriculture, cities, and the state in
the Near East. London.
Mallowan, M.E.L. 1936: The excavations at Tall Chagar Bazar, and an archaeological survey of the
Habur region, 1934–5. Iraq 3/1, 1–86.
—. 1946: Excavations in the Balikh Valley, 1938. Iraq 8, 111–62.
Manning, S.W., M. Barbetti, B. Kromer, P.I. Kuniholm, I. Levin, M.W. Newton, and P.J. Reimer 2002:
No systematic early bias to Mediterranean 14C ages: radiocarbon measurements from tree-ring and
air samples provide tight limits to age offsets. Radiocarbon 44/3, 739–54.
Manning, S.W., B. Kromer, P.I. Kuniholm, and M.W. Newton 2001: Anatolian tree rings and a new
chronology for the East Mediterranean Bronze-Iron Ages. Science 294/5551, 2532–5.
163
Manning, S.W., C.B. Ramsey, W. Kutschera, T. Higham, B. Kromer, P. Steier, and E.M. Wild 2006:
Chronology for the Aegean Late Bronze Age 1700–1400 B.C. Science 312/5773, 565–9.
Marchesi, G. 2006: LUMMA in the onomasticon and literature of ancient Mesopotamia. History of the
Ancient Near East/Studies, 10. Padova.
Marfoe, L. 1979: The integrative transformation: patterns of sociopolitical organization in southern
Syria. Bulletin of the American Schools of Oriental Research 234, 1–42.
Margueron, J.-C. 1991: Mari, l’Euphrate, et le Khabur au milieu du IIIe millénaire. Bulletin of the
Canadian Society for Mesopotamian Studies 21, 79–100.
Marro, C. 2004a: Itinéraires et voies de circulation du Caucase à l’Euphrate: le rôle des nomades dans
le système d’échanges et l’économie protohistorique des IVème–IIIème millénaires avant notre ère.
In C. Nicolle (ed.): Nomades et sédentaires dans le Proche Orient ancien. Amurru, 3. Paris, 51–62.
—. 2004b: Upper Mesopotamia and the Caucasus: an essay on the evolution of routes and road
networks from the Old Assyrian kingdom to the Ottoman empire. In A. Sagona (ed.): A view
from the highlands: Archaeological studies in honour of Charles Burney. Ancient Near Eastern studies.
Supplement, 12. Leuven, 91–120.
—. 2007a: Upper-Mesopotamia and the late third millennium crisis hypothesis: state of the art and
issues at stake. In C. Kuzucuoğlu and C. Marro (eds.): Sociétés humaines et changement climatique à la
13–20.
—. 2007b: Continuity and change in the Birecik Valley at the end of the third millennium B.C:
the archaeological evidence from Horum Höyük. In C. Kuzucuoğlu and C. Marro (eds.):
Marro, C., A. Tibet, and F. Bulgan 2000: Fouilles de sauvetage de Horum Höyük (province de
Gaziantep): Quatrième rapport préliminaire. Anatolia Antiqua 8, 257–78.
Marro, C., A. Tibet, and R. Ergeç 1997: Fouilles de sauvetage de Horum Höyük (province de
Gaziantep): Premier rapport préliminaire. Anatolia Antiqua 5, 371–91.
—. 1998: Fouilles de sauvetage de Horum Höyük (province de Gaziantep): Troisième rapport
préliminaire. Anatolia Antiqua 7, 285–307.
Matthews, V.H. 1978: Pastoral nomadism in the Mari Kingdom (ca. 1830–1760 B.C.). American Schools of
Oriental Research Dissertation Series, 3. Cambridge, Mass.
Matthews, W. 2003: Microstratigraphic sequences: Indications of uses and concepts of space. In R.
Matthews (ed.): Excavations at Tell Brak, 4. Exploring an upper Mesopotamian regional centre, 1994–1996.
Mayewski, P.A., E.E. Rohling, J.C. Stager, W. Karlén, K.A. Maasch, L.D. Meeker, E.A. Meyerson, F.
Gasse, S. van Kreveld, K. Holmgren, J. Lee-Thorp, G. Rosqvist, F. Rack, M. Staubwasser, R.R.
Schneider, and E.J. Steig 2004: Holocene climate variability. Quaternary Research 62/3, 243–55.
McCorriston, J. 1992: The Halaf environment and human activities in the Khabur Drainage, Syria.
Journal of Field Archaeology 19/3, 315–33.
—. 1997: The fiber revolution. Textile extensification, alienation, and social stratification in ancient
Mesopotamia. Current Anthropology 38/4, 517–49.
McIntosh, R.J., J.A. Tainter, and S.K. McIntosh 2000a: Climate, history, and human action. In R.J.
McIntosh, J.A. Tainter, and S.K. McIntosh (eds.): The way the wind blows: climate, history, and human
action. Historical Ecology Series. New York, 1–42.
— (eds.) 2000b: The way the wind blows: climate, history, and human action. Historical Ecology Series. New
York.
McLellan, T.L., R. Grayson, and C. Oglesby 2000: Bronze Age water harvesting in north Syria. In O.
Rouault and M. Wäfler (eds.): La Djéziré et l’Euphrate syriens de la protohistoire à la fin du IIe millénaire
av. J.-C.: tendances dans l’interprétation historique des données nouvelles. Subartu, 7. Turnhout, 137–55.
164
Bibliography
McLellan, T.L. and A. Porter 1995: Jawa and North Syria. Studies in the History and Archaeology of Jordan
5, 49–65.
Meijer, D.J.W. 1986: A survey in Northeastern Syria. Uitgaven van het Nederlands HistorischArchaeologisch Instituut te Istanbul, 58. Istanbul.
—. 1989: Hammam al-Turkman on the Balikh: First results of the fifth campaign, 1988. Akkadica
64–65, 1–12.
—. 1990: An archaeological survey: some assumptions and ideas. In S.M. Eichler, M. Wäfler, and D.
Warburton (eds.): Tall al-Hamidiyah 2. Orbis Biblicus et Orientalis, 6. Freiburg, 31–45.
—. 1996: Tell Hammam al-Turkman: preliminary report on the seventh campaign, May–July 1995.
Anatolica 22, 181–93.
—. 2000: Ecology and archaeology: perceptions and questions. In L. Milano, S. de Martino, F.M. Fales,
and G.B. Lanfranchi (eds.): Landscapes: territories, frontiers and horizons in the ancient Near East. History
of the Ancient Near East, 3. Padova, 203–12.
—. 2001: Architecture and stratigraphy. In M.N. Van Loon (ed.): Selenkahiye. Final report on the University
of Chicago and University of Amsterdam excavations in the Tabqa Reservoir, northern Syria, 1967–1975.
Uitgaven van het Nederlands Historisch-Archaeologisch Instituut te Istanbul, 91. Istanbul, 25–125.
—. 2007: The area of the Balikh between ca. 2500 and 1700 BC. In P. Matthiae, F. Pinnock, L. Nigro,
and L. Peyronel (eds.): Proceedings of the international colloquium: From relative chronology to absolute
chronology: The second millennium BC in Syria–Palestine (Rome, 29th November–1st December 2001).
Contributi del Centro Linceo Interdisciplinare «Beniamino Segre», 117. Rome, 313–26.
—. forthcoming: Nomadism, pastoralism, and town and country: about the roaming elements in the
Syrian Middle Bronze Age. In P.R. de Miroschedji (ed.): Proceedings of 3ICAANE (Paris, 2002).
Meyer, J.-W. 1997: Djebelet el-Beda: eine Stätte der Ahnenverehrung? Altorientalische Forschungen 24/2,
294–309.
—. 2006: Zur Frage der Urbanisierung von Tell Chuera. In P. Butterlin, M. Lebeau, J.-Y. Monchambert,
J.-L. Montero Fenollós, and B. Muller (eds.): Les espaces syro-mésopotamiens. Dimensions de l’expérience
humaine au Proche-Orient ancien. Volume d’hommage offert à Jean-Claude Margueron. Subartu, 17.
Turnhout, 179–89.
Michalowski, P. 1976: The royal correspondence of Ur. Dissertation, Yale University.
—. 2005: Literary works from the court of king Ishbi-Erra of Isin. In Y. Sefati, P. Artzi, C. Cohen,
B.L. Eichler, and V.A. Hurowitz (eds.): “An experienced scribe who neglects nothing”: ancient Near Eastern
studies in honor of Jacob Klein. Bethesda, 199–211.
Miglus, P.A. and E. Strommenger 2002: Stadtbefestigungen, Häuser und Tempel. Ausgrabungen in Tall Bi’a/
Tuttul, 8. Saarbrücken.
Migowski, C., M. Stein, S. Prasad, J.F.W. Negendank, and A. Agnon 2006: Holocene climate variability
and cultural evolution in the Near East from the Dead Sea sedimentary record. Quaternary Research
66/3, 421–31.
Millet Albà, A. 2004: La localisation des terroirs benjaminites du royaume de Mari. In C. Nicolle (ed.):
Minc, L.D. and K.P. Smith 1989: The spirit of survival: cultural responses to resource variability in
North Alaska. In P. Halstead and J. O’Shea (eds.): Bad year economics: cultural responses to risk and
uncertainty. New Directions in Archaeology. Cambridge, 8–39.
Minnis, P.E. 1996: Notes on economic uncertainty and human behavior in the prehistoric North
American Southwest. In J.A. Tainter and B.B. Tainter (eds.): Evolving complexity and environmental risk
in the prehistoric Southwest. Santa Fe Institute Studies in the Sciences of Complexity Proceedings, 24.
Reading, 57–78.
Moore, A.M.T., G. Hillman, C., and A.J. Legge (eds.) 2000: Village on the Euphrates. From foraging to
farming at Abu Hureyra. Oxford.
165
Moortgat-Correns, U. 1972: Die Bildwerke von Djebelet el Bēdā in ihrer räuumlichen und zeitlichen Umwelt.
Berlin.
Mulders, M.A. 1969: The arid soils of the Balikh Basin (Syria). Unpublished dissertation, Universiteit van
Utrecht.
Neely, J.A. and H.T. Wright 1994: Early settlement and irrigation on the Deh Luran Plain. Village and early
state societies in Southwestern Iran. University of Michigan Museum of Anthropology Technical
Report, 26. Ann Arbor.
Nettle, D. 1996: Language diversity in West Africa: an ecological approach. Journal of Anthropological
Archaeology 15/4, 403–38.
—. 1998: Explaining global patterns of language diversity. Journal of Anthropological Archaeology 17/4,
354–74.
Neumann, F., C. Schölzel, T. Litt, A. Hense, and M. Stein 2006: Holocene vegetation and climate
history of the northern Golan heights (Near East). Vegetation History and Archaeobotany 16, 329–46.
Nicolle, C. 2006: Tell Mohammed Diyab 3. Travaux de 1992–2000 sur les buttes A et B. Paris.
Oates, D. and J. Oates 2001: Archaeological reconstruction and historical commentary. In D. Oates,
J. Oates, and H. McDonald (eds.): Excavations at Tell Brak, 2. Nagar in the third millennium BC.
Oates, D., J. Oates, and H. McDonald (eds.) 2001: Excavations at Tell Brak, 2. Nagar in the third millennium
BC. Cambridge.
Oates, J. 2004: Archaeology in Mesopotamia: digging deeper at Tell Brak (Albert Reckitt Archaeological
Lecture, British Academy, 2004). Available at: http://www.mcdonald.cam.ac.uk/projects/brak/
index.htm (Accessed 7 November, 2008).
Ökse, A.T. 2005: Gre Virike: A ritual centre for Early Bronze Age rural communities on the Middle
Euphrates. Antiquity 79/306, http://antiquity.ac.uk/ProjGall/okse/index.html (Accessed 17
March, 2009).
—. 2006: Gre Virike (period I) Early Bronze Age ritual facilities on the Middle Euphrates River.
Anatolica 32, 1–27.
—. 2007: A ‘high’ terrace at Gre Virike to the north of Carchemish: power of local rulers as founders?
In E. Peltenburg (ed.): Euphrates river valley settlement: the Carchemish sector in the third millennium BC.
Levant Supplementary Series, 5. Oxford, 94–104.
Ökse, A.T., A. Engin, V.M. Tekinalp, H.U. Dağ, and A. Görüş 2001: Research at Mezraa Höyük, 1999.
In N. Tuna, J. Öztürk, and J. Velibeyoğlu (eds.): Salvage Project of the Archaeological Heritage of the Ilısu
and Carchemish Dam Reservoirs. Activities in 1999. Ankara, 213–32.
Ökse, A.T. and V.M. Tekinalp 1999: Mezraa Höyük: research in 1998. In N. Tuna and J. Öztürk (eds.):
Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs. Activities in 1998
Ankara, 199–214.
Osborne, R. 2004: Demography and survey. In S.E. Alcock and J.F. Cherry (eds.): Side-by-side survey:
comparative regional studies in the Mediterranean world. Oxford, 163–172.
Otto, A. 2006: Alltag und Gesellschaft zur Spätbronzezeit: eine Fallstudie aus Tall Bazi (Syrien). Subartu, 19.
Turnhout.
Parkinson, W.A. 2002: Introduction: archaeology and tribal societies. In W.A. Parkinson (ed.): The
archaeology of tribal societies. Archaeological Series, 15. Ann Arbor, 1–12.
Pearson, J.A., H. Buitenhuis, R.E.M. Hedges, L. Martin, N. Russell, and K.C. Twiss 2007: New light on
early caprine herding strategies from isotope analysis: a case study from Neolithic Anatolia. Journal
of Archaeological Science 34/12, 2170–9.
Peltenburg, E. 2000: From nucleation to dispersal. Late third millennium BC settlement pattern
transformation in the Near East and Aegean. In O. Rouault and M. Wäfler (eds.): La Djéziré et
l’Euphrate syriens de la protohistoire à la fin du IIe millénaire av. J.-C.: tendances dans l’interprétation historique
des données nouvelles. Subartu, 7. Turnhout, 183–206.
166
Bibliography
—. 2007a: New perspectives on the Carchemish sector of the Middle Euphrates River valley in the 3rd
millennium BC. In E. Peltenburg (ed.): Euphrates river valley settlement: the Carchemish sector in the third
millennium BC. Levant Supplementary Series, 5. Oxford, 1–24.
—. 2007b: Diverse settlement pattern changes in the middle Euphrates Valley in the later third
millennium B.C: the contribution of Jerablus Tahtani. In C. Kuzucuoğlu and C. Marro (eds.):
— (ed.) 2007c: Euphrates river valley settlement: the Carchemish sector in the third millennium BC. Levant
Supplementary Series, 5. Oxford.
—. forthcoming: The emergence of Carchemish as a major polity: contributions from the Land of
Carchemish Project (Syria), 2006. In: Proceedings of 6ICAANE (Rome, 2008).
Peters, J., D. Helmer, A. von den Driesch, and M. Saña Segui 1999: Early animal husbandry in the
Northern Levant. Paléorient 25/2, 27–48.
Peterson, C.E. and R.D. Drennan 2005: Communities, settlements, sites, and surveys: regional-scale
analysis of prehistoric human interaction. American Antiquity 70/1, 5–30.
Petit, L.P. forthcoming: Settlement dynamics in the Middle Jordan Valley during Iron II
Pfälzner, P. 1998: Eine Modifikation der Periodisierung Nordmesopotamiens im 3. Jtsd. v. Chr.
Mitteilungen der Deutschen Orient-Gesellschaft zu Berlin 130, 69–71.
—. 2001: Haus und Haushalt: Wohnformen des dritten Jahrtausends vor Christus in Nordmesopotamien.
Damaszener Forschungen, 9. Mainz am Rhein.
Pfälzner, P. and A. Wissing 2004: Urbanismus in der Unterstadt von Urkeš. Ergebnisse einer
geomagnetischen Prospektion und eines archäologischen Surveys in der südostlichen Unterstadt
von Tall Mozan im Sommer 2002. Mitteilungen der Deutschen Orient-Gesellschaft zu Berlin 136, 41–86.
Plog, S. and J.L. Hantman 1990: Chronology construction and the study of prehistoric culture change.
Poidebard, A. 1934: La trace de Rome dans le désert de Syrie: le Limes de Trajan à la conquète arabe. Recherches
aériennes (1925–1932). Paris.
Pollock, S. 1999: Ancient Mesopotamia: the Eden that never was. Case Studies in Early Societies. Cambridge.
Porter, A. 2002: The dynamics of death: ancestors, pastoralism, and the origins of a third-millennium
city in Syria. Bulletin of the American Schools of Oriental Research 325, 1–36.
—. 2004: The urban nomad: countering old clichés. In C. Nicolle (ed.): Nomades et sédentaires dans le
Proche Orient ancien. Amurru, 3. Paris, 69–74.
—. 2007: You say potato, I say... typology, chronology and the origins of the Amorites. In C.
Kuzucuoğlu and C. Marro (eds.): Sociétés humaines et changement climatique à la fin du troisième millénaire:
une crise a-t-elle eu lieu en Haute Mésopotamie? Varia Anatolica, 19. Istanbul, 69–115.
—. 2009: Beyond dimorphism: ideologies and materialities of kinship as time-space distanciation. In
J. Szuchman (ed.): Nomads, tribes, and the state in the ancient Near East. Oriental Institute Seminars, 5.
Chicago, 201–25.
Postgate, N. 1994: How many Sumerians per hectare? – Probing the anatomy of an early city. Cambridge
Archaeological Journal 4/1, 47–65.
Powell, M.A. 1990: Maße und Gewichte. In D.O. Edzard (ed.): Reallexikon der Assyriologie und
Vorderasiatische Archäologie, VII. Berlin.
Pruß, A. 2004: Remarks on the chronological periods. In S. Anastasio, M. Lebeau, and M. Sauvage:
Atlas of preclassical Upper Mesopotamia. Subartu, 13. Turnhout, 7–21.
Pustovoytov, K., K. Schmidt, and H. Taubald 2007: Evidence for Holocene environmental changes in
the northern Fertile Crescent provided by pedogenic carbonate coatings. Quaternary Research 67/3,
315–27.
Rappaport, R.A. 1979: Ecology, meaning, and religion. Berkeley.
167
Rautman, A.E. 1996: Risk, reciprocity, and the operation of social networks. In J.A. Tainter and B.B.
Tainter (eds.): Evolving complexity and environmental risk in the prehistoric Southwest. Santa Fe Institute
Studies in the Sciences of Complexity Proceedings, 24. Reading, 197–222.
Read, D.W. and S.A. LeBlanc 2003: Population growth, carrying capacity, and conflict. Current
Reade, J. 2001: Assyrian king-lists, the royal tombs of Ur, and Indus origins. Journal of Near Eastern
Studies 60/1, 1–19.
Redding, R.W. 1993: Subsistence security as a selective pressure favoring increasing cultural complexity.
Bulletin on Sumerian Agriculture 7, 77–98.
Redman, C.L. 1982: Archaeological survey and the study of Mesopotamian urban systems. Journal of
Redman, C.L. and P.J. Watson 1970: Systematic, intensive surface collection. American Antiquity 35/3,
279–91.
Reichel, C.D. 2004: Appendix B: site gazetteer. In T.J. Wilkinson: On the margin of the Euphrates.
Settlement and land use at Tell es-Sweyhat and in the upper Lake Assad area, Syria. Oriental Institute
Publications, 124. Chicago, 335–48.
Reifenberg, A. 1952: The soils of Syria and the Lebanon. Journal of Soil Science 3/1, 68–88.
Renfrew, C. 1972: The emergence of civilisation: the Cyclades and the Aegean in the third millennium B.C. Studies
in Prehistory. London.
Renfrew, C. and P. Bahn 1996 (1991): Archaeology. Theories, methods and practice. London.
Riehl, S. 2008: Climate and agriculture in the ancient Near East: a synthesis of the archaeobotanical
and stable carbon isotope evidence. Vegetation History and Archaeobotany 17/Supplement 1, 43–51.
—. 2009: Archaeobotanical evidence for the interrelationship of agricultural decision-making and
climate change in the ancient Near East. Quaternary International 197/1–2, 93–114.
Riehl, S., R.A. Bryson, and K. Pustovoytov 2008: Changing growing conditions for crops during
the Near Eastern Bronze Age (3000–1200 BC): The stable carbon isotope evidence. Journal of
Archaeological Science 35/4, 1011–22.
Ristvet, L. 2005: Settlement, economy, and society in the Tell Leilan Region, Syria, 3000–1000 BC. Unpublished
dissertation, University of Cambridge.
—. 2008: Legal and archaeological territories of the second millennium BC in northern Mesopotamia.
Antiquity 82/317, 585–99.
Ristvet, L. and H. Weiss 2005: The Habur region in the late third and early second millennium BC. In
W. Orthmann (ed.): The history and archaeology of Syria, 1. Saarbrucken.
Roberts, B.K. 1996: Landscapes of settlement. Prehistory to the present. London.
Röllig, W. and H. Kühne 1977/8: The Lower Habur. A preliminary report on a survey conducted by
the Tubinger Atlas des Vorderen Orients in 1975. Annales Archéologiques Arabes Syriennes 27–28,
115–40.
—. 1983: The Lower Habur second preliminary report on a survey in 1977. Annales Archéologiques
Arabes Syriennes 33/2, 187–99.
Rosen, A.M. 1995: The social response to environmental change in Early Bronze Age Canaan. Journal
of Anthropological Archaeology 14, 26–44.
—. 1997: The geoarchaeology of Holocene environments and land use at Kazane Hoyuk, S.E. Turkey.
Geoarchaeology 12/4, 395–416.
—. 2007: Civilizing climate. Social responses to climate change in the ancient Near East. Lanham.
Rosen, A.M. and S.A. Rosen 2001: Determinist or not determinist?: climate, environment, and
archaeological explanation in the Levant. In S.R. Wolff (ed.): Studies in the archaeology of Israel and
neighboring lands: in memory of Douglas L. Esse. ASOR Books, 5. Chicago, 535–49.
168
Bibliography
Rosen, S.A. 2003: Early multi-resource nomadism: excavations at the Camel Site in the central Negev.
Antiquity 77/298, 749–61.
—. 2008: Desert pastoral nomadism in the longue durée. A case study from the Negev and the Southern
Levantine deserts. In H. Barnard and W. Wendrich (eds.): The archaeology of mobility. Old World and
New World nomadism. Cotsen Advanced Seminar Series, 4. Los Angeles, 115–40.
Rothman, M.S., R. Ergeç, N.F. Miller, J.A. Weber, and G. Kozbe 1998: Yarim Höyük and the Uruk
expansion (Part I). Anatolica 24, 65–99.
Rothman, M.S. and J.G. Fuensanta 2003: The archaeology of the Early Bronze I and II periods in
southeastern Turkey and north Syria. In M. Özdoğan, H. Hauptmann, and N. Başgelen (eds.): From
village to cities. Early villages in the Near East. Studies presented to Ufuk Esin. Istanbul, 583–622.
Rowe, A. 2002: Structural responses to resource stress: evidence for the evolution of social and
economic associations from Jordanian pastoral society. In W. Wendrich and G. Van der Kooij
(eds.): Moving matters. Ethnoarchaeology in the Near East. Proceedings of the international seminar held at
Cairo, 7–10 december 1998. CNWS Publications, 111. Leiden, 29–43.
Rowton, M.B. 1973a: Urban autonomy in a nomadic environment. Journal of Near Eastern Studies
32/1/2, 201–15.
—. 1973b: Autonomy and nomadism in Western Asia. Orientalia 42, 247–258.
—. 1974: Enclosed nomadism. Journal of the Economic and Social History of the Orient 17, 1–30.
—. 1980: Pastoralism and the periphery in evolutionary perspective. In M.-T. Barrelet (ed.): L’archéologie
de l’Iraq du début de l’époque neólithique à 333 avant notre ère: perspectives et limites de l’interprétation
anthropologique des documents. Colloques Internationaux du Centre National de la Recherche
Scientifique, 580. Paris.
Russell, K.W. 1988: After Eden. The behavioral ecology of early food production in the Near East and North
Africa. BAR International Series, 391. Oxford.
Rutishauser, S. 2008: Hollow ways in der Fernerkundung. In T. Hofmeier and O. Kaelin (eds.):
Stückwerk. Beiträge zur Kulturgeschichte des Alten Orients. Berlin, 215–40.
Sachau, E. 1883: Reise in Syrien und Mesopotamien. Leipzig.
Sahlins, M. 1972: Stone age economics. New York.
Said, R. 1993: The river Nile. Geology, hydrology and utilization. Oxford.
Sallaberger, W. 1996: Grain accounts: personnel lists and expenditure documents. In F. Ismail, W.
Sallaberger, P. Talon, and K. van Lerberghe (eds.): Administrative documents from Tell Beydar (seasons
1993–1995). Subartu, 2. Turnhout, 81–106.
—. 1998: Ein Synchronismus der Urkunden von Tell Beydar mit Mari und Ebla? In M. Lebeau (ed.):
About Subartu: studies devoted to Upper Mesopotamia. Subartu, 4/2. Turnhout, 23–39.
—. 2004a: Relative Chronologie von der späten frühdynastischen bis zur altbabylonischen Zeit. In J.-W.
Meyer and W. Sommerfeld (eds.): 2000 v. Chr.: politische, wirtschaftliche und kulturelle Entwicklung im
Zeichen einer Jahrtausendwende. Colloquien der Deutschen Orient-Gesellschaft, 3. Saarbrücken, 15–43.
—. 2004b: A note on the sheep and goat flocks. Introduction to texts 141–167. In L. Milano, W.
Sallaberger, P. Talon, and K. van Lerberghe (eds.): Third millennium cuneiform texts from Tell Beydar
(seasons 1996–2002). Subartu, 12. Turnhout, 13–21.
—. 2007: From urban culture to nomadism: A history of Upper Mesopotamia in the late third
millennium. In C. Kuzucuoğlu and C. Marro (eds.): Sociétés humaines et changement climatique à la fin du
troisième millénaire: une crise a-t-elle eu lieu en Haute Mésopotamie? Varia Anatolica, 19. Istanbul, 417–56.
Sallaberger, W. and P. Talon 1996: Transliterated texts. In F. Ismail, W. Sallaberger, P. Talon, and K. van
Lerberghe (eds.): Administrative documents from Tell Beydar (seasons 1993–1995). Subartu, 2. Turnhout,
127–74.
Salzman, P.C. 1978a: Ideology and change in Middle Eastern tribal societies. Man 13/4, 618–37.
—. 1978b: Does complementary opposition exist? American Anthropologist 80/1, 53–70.
169
Savage, S.H. 1997: Assessing departures from log-normality in the rank-size rule. Journal of
Archaeological Science 24/3, 233–44.
Schacht, R.M. 1984: The contemporaneity problem. American Antiquity 49/4, 678–95.
Schiffer, M.B. 1987: Formation processes in the archaeological record. Albuquerque.
Schirmer, W. 1987: Landschaftsgeschichte um Tall Bi’a am syrischen Euphrat. Mitteilungen der Deutschen
Schwartz, G.M. 1994: Rural economic specialization and early urbanization in the Khabur Valley, Syria.
In G.M. Schwartz and S.E. Falconer (eds.): Archaeological views from the countryside: village communities in
early complex societies. Washington, 19–36.
—. 2007: Taking the long view on collapse: A Syrian perspective. In C. Kuzucuoğlu and C. Marro
(eds.): Sociétés humaines et changement climatique à la fin du troisième millénaire: une crise a-t-elle eu lieu en
Haute Mésopotamie? Varia Anatolica, 19. Istanbul, 45–67.
Schwartz, G.M. and H.H. Curvers 1992: Tell al-Raqa’i 1989 and 1990: Further investigations at a small
rural site of early urban northern Mesopotamia. American Journal of Archaeology 96/3, 397–419.
Schwartz, G.M., H.H. Curvers, F.A. Gerritsen, J.A. MacCormack, N.F. Miller, and J.A. Weber 2000:
Excavation and survey in the Jabbul Plain, Western Syria: the Umm el-Marra project 1996–1997.
American Journal of Archaeology 104/3, 419–62.
Sertok, K. 2007: Fruit stands and the definition of a cultural area around Carchemish. In E.
Peltenburg (ed.): Euphrates river valley settlement: the Carchemish sector in the third millennium BC. Levant
Supplementary Series, 5. Oxford, 238–49.
Sertok, K. and R. Ergeç 1999a: A new Early Bronze Age cemetery: excavations near the Birecik Dam,
southeastern Turkey. Anatolica 25, 87–107.
—. 1999b: Seraga Höyük: Research in 1998. In N. Tuna and J. Öztürk (eds.): Salvage Project of the
Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs. Activities in 1998 Ankara, 168–74.
Sertok, K. and F. Kulakoğlu 2001: Results of the 1999 season excavations at Şaraga Höyük. In N.
—. 2002: Şaraga Höyük 2000. In N. Tuna, J. Öztürk, and J. Velibeyoğlu (eds.): Salvage Project of the
Archaeological Heritage of the Ilısu and Carchemish Dam Reservoirs. Activities in 2000. Ankara, 370–81.
Sertok, K., F. Kulakoğlu, and F. Squadrone 2004: Şaraga Höyük 2001. In N. Tuna, J. Öztürk, and
J. Velibeyoğlu (eds.): Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish Dam
Reservoirs. Activities in 2001. Ankara, 308–15.
—. 2007: Living along and together with the Euphrates. The effects of the Euphrates on a long-life
settlement as Şaraga Höyük. In C. Kuzucuoğlu and C. Marro (eds.): Sociétés humaines et changement
climatique à la fin du troisième millénaire: une crise a-t-elle eu lieu en Haute Mésopotamie? Varia Anatolica, 19.
Istanbul, 341–53.
Sherratt, A. 1981: Plough and pastoralism: aspects of the secondary products revolution. In I. Hodder,
G. Isaac, and N. Hammond (eds.): Pattern of the past: studies in honour of David Clark. Cambridge,
261–305.
Sołtysiak, A. forthcoming: Human remains from Tell Barri, area G. In R. Pierobon-Benoit (ed.): Tell
Barri/Kahat 3.
Sommerfeld, W. 2000: Narām-Sîn, die “Große Revolte” und MAR.TUki. In J. Marzahn and H.
Neumann (eds.): Assyriologica et Semitica. Festschrift für Joachim Oelsner anläßlich seines 65. Geburtstages am
18. Februar 1997. Alter Orient und Altes Testament, 252. Münster, 419–36.
Staubwasser, M. and H. Weiss 2006: Holocene climate and cultural evolution in late prehistoric-early
historic West Asia. Quaternary Research 66/3, 372–87.
Stein, G.J. 1987: Regional economic integration in early state societies: third millennium B.C. pastoral
production in Gritille, southeast Turkey. Paléorient 13/2, 101–11.
170
Bibliography
Stein, G.J., R. Bernbeck, C. Coursey, A. McMahon, N.F. Miller, A. Misir, J. Nicola, H. Pittman, S.
Pollock, and H.T. Wright 1996: Uruk colonies and Anatolian communities: an interim report on
the 1992–1993 excavations at Hacinebi, Turkey. American Journal of Archaeology 100/2, 205–60.
Stein, G.J. and P. Wattenmaker 1990: The 1987 Tell Leilan regional survey: preliminary report. In N.F.
Miller (ed.): Economy and settlement in the Near East: analyses of ancient sites and materials. MASCA
Research Papers in Science and Archaeology. Supplement, 7. Philadephia, 8–18.
Steinkeller, P. 1992: Early Semitic literature and third millennium seals with mythological motifs. In P.
Fronzaroli (ed.): Literature and literary language at Ebla. Quaderni di Semitistica, 18. Firenze, 243–83.
—. 1998: The historical background of Urkesh and the Hurrian beginnings in northern Mesopotamia.
In G. Buccellati and M. Kelly-Buccellati (eds.): Urkesh and the Hurrians. Studies in honor of Lloyd
Cotsen Bibliotheca Mesopotamica, 26. Malibu, 75–98.
Stevens, L.R., E. Ito, A. Schwalb, and H.E. Wright Jr. 2006: Timing of atmospheric precipitation in the
Zagros Mountains inferred from a multi-proxy record from Lake Mirabad, Iran. Quaternary Research
66/3, 494–500.
Stevens, L.R., H.E. Wright Jr., and E. Ito 2001: Proposed changes in seasonality of climate during the
Lateglacial and Holocene at Lake Zeribar, Iran. The Holocene 11/6, 747–55.
Stol, M. 1995: Old Babylonian cattle. Bulletin on Sumerian Agriculture 8, 173–213.
Stone, E.C. 1981: Texts, architecture and ethnographic analogy: patterns of residence in Old
Babylonian Nippur. Iraq 43, 19–33.
—. 1987: Nippur neighborhoods. Studies in Ancient Oriental Civilization, 44. Chicago.
Stone, G.D. 1994: Agricultural intensification and perimetrics: ethnoarchaeological evidence from
Nigeria. Current Anthropology 35/3, 317–24.
Stone, T. 2003: Social identity and ethnic interaction in the Western Pueblos of the American
Southwest. Journal of Archaeological Method and Theory 10/1, 31–67.
Stordeur, D. 1993: Sédentaires et nomades du PPNB final dans le désert de Palmyre (Syrie). Paléorient
19/1, 187–204.
Stothers, R.B. 1999: Volcanic dry fogs, climate cooling, and plague pandemics in Europe and the
Middle East. Climatic Change 42, 713–23.
Streck, M.P. 2000: Das amurritische Onomastikon der altbabylonischen Zeit I: die Amurriter, die onomastische
Forschung, Orthographie und Phonologie, Nominalphonologie. Alter Orient und Altes Testament, 271/1.
Münster.
—. 2002: Zwischen Weide, Dorf und Stadt: sozio-ökonomische Strukturen des amurritischen
Nomadismus am Mittleren Euphrat. Baghdader Mitteilungen 33, 155–209.
—. 2004: Die Amurriter der altbaylonischen Zeit im Spiegel des Onomastikons. Eine ethnolinguistische Evaluierung. In J.-W. Meyer and W. Sommerfeld (eds.): 2000 v. Chr.: politische,
wirtschaftliche und kulturelle Entwicklung im Zeichen einer Jahrtausendwend. Colloquien der Deutschen
Orient-Gesellschaft, 3. Saarbrücken, 313–55.
Strommenger, E. 1981: Die archäologischen Forschungen in Tall Bi’a 1980. Mitteilungen der Deutschen
Sumner, W.M. 1979: Estimating population by analogy: an example. In C. Kramer (ed.):
Ethnoarchaeology: Implications of ethnography for archaeology. New York, 164–74.
—. 1990: An archaeological estimate of population trends since 6000 BC in the Kur River Basin, Fars
Province, Iran. In M. Taddei (ed.): South Asian archaeology 1987: proceedings of the ninth International
Conference of the Association of South Asian Archaeologists in Western Europe, held in the Fondazione Giorgio
Cini, Island of San Giorgio Maggiore, Venice. Serie orientale Roma, 66. Rome, 3–16.
Sykes, M. 1915: The Caliphs’ last heritage. A short history of the Turkish empire. London.
Szarzynska, K. 2002: Sheep husbandry and production of wool, garments and cloths in archaic Sumer. Warsaw.
Tainter, J.A. 1988: The collapse of complex societies. New Studies in Archaeology. Cambridge.
171
Tainter, J.A. and B.B. Tainter (eds.) 1996: Evolving complexity and environmental risk in the prehistoric
Southwest. Santa Fe Institute Studies in the Sciences of Complexity Proceedings, 24. Reading.
Talon, P. 1996: Personal names. In F. Ismail, W. Sallaberger, P. Talon, and K. van Lerberghe (eds.):
Administrative documents from Tell Beydar (seasons 1993–1995). Subartu, 2. Turnhout, 75–80.
Taşkıran, H. 2002: The Palaeolithic survey in the Carchemish Dam reservoir region: 2000 season. In N.
Thalen, D.C.P. 1979: Ecology and utilization of desert shrub rangelands in Iraq. The Hague.
Thissen, L. 1989: An Early Bronze III pottery region between the Middle Euphrates and Habur:
new evidence from Tell Hammam et-Turkman. In O.M.C. Haex, H.H. Curvers, and P.M.M.G.
Akkermans (eds.): To the Euphrates and beyond. Archaeological studies in honour of Maurits N. van Loon.
Rotterdam, 195–211.
Ur, J.A. 2003: CORONA satellite photography and ancient road networks: a northern Mesopotamian
case study. Antiquity 77/295, 102–15.
—. 2004: Urbanism and society in the third millennium Upper Khabur Basin. Unpublished dissertation,
University of Chicago.
van de Mieroop, M. 1997: The ancient Mesopotamian city. Oxford.
—. 2007: A history of the ancient Near East, ca. 3000–323 BC. Second edition. Blackwell History of the
Ancient World. Malden.
van der Kooij, G. 2001: The vicissitudes of life at Dayr ‘Allā during the first millennium BC, seen in
a wider context. In G. Bisheh (ed.): Studies in the history and archaeology of Jordan VII: Jordan by the
millenia. Amman, 295–303.
van der Steen, E. 2009: Tribal societies in the nineteenth century: a model. In J. Szuchman (ed.):
Nomads, tribes, and the state in the ancient Near East. Oriental Institute Seminars, 5. Chicago, 105–17.
van Driel, G. 2005: Ethnicity, how to cope with the subject. In W.H. van Soldt, R. Kalvelagen, and D.
Katz (eds.): Ethnicity in ancient Mesopotamia: papers read at the 48th Rencontre Assyriologique Internationale,
Leiden, 1–4 July 2002. Uitgaven van het Nederlands Instituut voor het Nabĳe Oosten te Leiden,
102. Leiden, 1–10.
van Liere, W.J. 1963: Capitals and citadels of Bronze-Iron Age Syria in their relationship to land and
water. Les Annales Archéologiques de Syrie 13, 109–22.
van Liere, W.J. and J. Lauffray 1954: Nouvelle prospection archeologique dans la Haute Jezireh
Syrienne. Annales Archéologiques Arabes Syriennes 4/5, 129–48.
van Loon, M.N. (ed.) 1988: Hammam et-Turkman I. Report on the University of Amsterdam’s 1981–84
excavations in Syria. Uitgaven van het Nederlands Historisch-Archaeologisch Instituut te Istanbul,
63. Istanbul.
van Loon, M.N. and D.J.W. Meijer 1988: Foreword. In M.N. van Loon (ed.): Hammam et-Turkman I.
Report on the University of Amsterdam’s 1981–1984 excavations in Syria. Uitgaven van het Nederlands
Historisch-Archaeologisch Instituut te Istanbul, 63. Istanbul, xxv–xxix.
van Neer, W. and B. Decupere 2000: Faunal remains from Tell Beydar (seasons 1992–1997). In K. van
Lerberghe and G. Voet (eds.): Tell Beydar: Environmental and technical studies. Subartu, 6. Turnhout,
69–115.
van Zeist, W. and S. Bottema 1977: Palynological investigations in western Iran. Palaeohistoria 19, 19–85.
van Zeist, W., W. Waterbolk-van Rooijen, and S. Bottema 1988: Some notes on the plant husbandry of
Tell Hammam et-Turkman. In M.N. van Loon (ed.): Hammam et-Turkman I. Report on the University
of Amsterdam’s 1981–84 excavations in Syria. Uitgaven van het Nederlands Historisch-Archaeologisch
Instituut te Istanbul, 63. Istanbul, 705–15.
van Zeist, W. and H. Woldring 1978: A postglacial pollen diagram from Lake Van in east Anatolia.
Review of Palaeobotany and Palynology 26, 249–76.
172
Bibliography
Vila, E. 1995: Analyse de la faune des secteurs nord et sud du Steinbau 1. In W. Orthmann, R.
Hempelmann, H. Klein, C. Kühne, M. Novak, A. Pruß, E. Vila, H.-M. Weicken, and A. Wener:
Ausgrabungen in Tell Chuēra in Nordost-Syrien I: Vorbericht über die Grabungskampagnen 1986 bis 1992.
Vorderasiatische Forschungen der Max Freiherr von Oppenheim-Stiftung, 2. Saarbrücken, 267–79.
Villard, P. 1987: Un conflit d’autorités à propos des eaux du Balih. MARI 5, 591–96.
von den Driesch, A. 1993: Faunal remains from Habuba Kabira in Syria. In H. Buitenhuis and A.T.
Clason (eds.): Archaeozoology of the Near East. Proceedings of the first international symposium on the
archaeozoology of southwestern Asia and adjacent areas. Leiden, 52–9.
Waetzoldt, H. 1996: Privathäuser. Ihre Grösse, Einrichtung und die Zahl der Bewohner. In K.R.
Veenhof (ed.): Houses and households in Ancient Mesopotamia. Uitgaven van het Nederlands HistorischArchaeologisch Instituut te Istanbul, 78. Istanbul, 145–52.
Wäfler, M. 2001: Tall al-Hamidiya 3. Zur historischen Geographie von Idamaras zur Zeit der Archive von Mari (2)
und Subat-enlil/Sehna. Orbis Biblicus et Orientalis, 21. Freiburg.
Wattenmaker, P. 1987: Town and village economies in an early state society. Paléorient 13/2, 113–22.
—. 1998: Household and state in Upper Mesopotamia: specialized economy and the social uses of goods in an early
complex society. Smithsonian Series in Archaeological Inquiry. Washington.
Weber, J.A. 1997: Faunal remains from Tell es-Sweyhat and Hajji Ibrahim. In R.L. Zettler: Subsistence
and settlement in a marginal environment: Tell es-Sweyhat, 1989–1995 preliminary report. MASCA Research
Papers in Science and Archaeology, 14. Philadelphia, 133–67.
Weiss, H. 1977: Periodization, population and early state formation in Khuzistan. In L.D. Levine and
T.C. Young Jr. (eds.): Mountains and lowlands: essays in the archaeology of greater Mesopotamia. Bibliotheca
Mesopotamica, 7. Malibu, 347–70.
—. 1986: The origins of Tell Leilan and the conquest of space in third millennium Mesopotamia.
In H. Weiss (ed.): The origins of cities in dry-farming Syria and Mesopotamia in the third millennium B.C.
Guilford, 71–108.
—. 1997: Leilan, Tell. In E.M. Meyers (ed.): The Oxford encyclopedia of archaeology in the Near East. Volume
3. New York, 341–7.
—. 2000: Beyond the Younger Dryas. Collapse as adaptation to abrupt climate change in ancient West
Asia and the Eastern Mediterranean. In G. Bawden and R.M. Reycraft (eds.): Environmental disaster
and the archaeology of human response. Anthropological Papers, 7. Albuquerque, 75–98.
—. 2007: Tell Leilan Project. Available at: http://leilan.yale.edu/index.html (Accessed 13 February,
2008).
Weiss, H. and R.S. Bradley 2001: What drives societal collapse? Science 291, 609–610.
Weiss, H., M.-A. Courty, W. Wetterstrom, F. Guichard, L.M. Senior, R. Meadow, and A. Curnow
1993: The genesis and collapse of third millennium north Mesopotamian civilization. Science 261,
995–1004.
Weiss, H., F. deLillis, D. deMoulins, J. Eidem, T. Guilderson, U. Kasten, T. Larsen, L. Mori, L. Ristvet,
E. Rova, and W. Wetterstrom 2002–3: Revising the contours of history at Tell Leilan. Annales
Archéologiques Arabes Syriennes 45–46, 59–74.
Wendrich, W. and H. Barnard 2008: The archaeology of mobility: definitions and research approaches.
In H. Barnard and W. Wendrich (eds.): The archaeology of mobility. Old World and New World nomadism.
Cotsen Advanced Seminar Series, 4. Los Angeles, 1–21.
Whallon, R. 2006: Social networks and information: non-“utilitarian” mobility among hunter-gatherers.
Journal of Anthropological Archaeology 25/2, 259–70.
Whiting, R.M. 1995: Amorite tribes and nations of second-millennium Western Asia. In J.M. Sasson
(ed.): Civilizations of the Ancient Near East. New York, 1231–42.
Wick, L., G. Lemcke, and M. Sturm 2003: Evidence of Lateglacial and Holocene climatic change and
human impact in eastern Anatolia: high-resolution pollen, charcoal, isotopic and geochemical
records from the laminated sediments of Lake Van, Turkey. The Holocene 13/5, 665–75.
173
Wilcke, C. 1969: Zur Geschichte der Amurriter in der Ur-III-Zeit. Welt des Orients 5, 1–31.
Wilkinson, T.J. 1989: Extensive sherd scatters and land-use intensity: some recent results. Journal of
—. 1990: The development of settlement in the north Jazira between the 7th and 1st millennia B.C.
Iraq 52, 49–62.
—. 1993: Linear hollows in the Jazira, Upper Mesopotamia. Antiquity 67, 548–62.
—. 1994: The structure and dynamics of dry-farming states in Upper Mesopotamia. Current
—. 1997: Environmental fluctuations, agricultural production and collapse: a view from Bronze Age
Upper Mesopotamia. In H.N. Dalfes, G. Kukla, and H. Weiss (eds.): Third millennium BC climate
change and Old World collapse. NATO ASI series, I/49. Berlin, 67–106.
—. 1998: Water and human settlement in the Balikh Valley, Syria: investigations from 1992–1995.
—. 1999: Demographic trends from archaeological survey: case studies from the Levant and the Near
East. In J. Bintliff and K. Sbonias (eds.): Reconstructing past population trends in Mediterranean Europe
(3000 BC–AD 1800). The Archaeology of Mediterranean Landscapes, 1. Oxford, 45–64.
—. 2000a: Regional approaches to Mesopotamian archaeology: The contribution of archaeological
surveys. Journal of Archaeological Research 8/3, 219–67.
—. 2000b: Settlement and land use in the zone of uncertainty in Upper Mesopotamia. In R.M. Jas
(ed.): Rainfall and agriculture in northern Mesopotamia. Uitgaven van het Nederlands HistorischArchaeologisch Instituut te Istanbul, 88. Istanbul, 1–35.
—. 2003: Archaeological landscapes of the Near East. Tucson.
—. 2004: On the margin of the Euphrates. Settlement and land use at Tell es-Sweyhat and in the upper Lake Assad
area, Syria. Oriental Institute Publications, 124. Chicago.
—. n.d.: Sabi Abyad: the geoarchaeology of a complex landscape. Manuscript.
Wilkinson, T.J., J.H. Christiansen, J.A. Ur, M. Widell, and M. Altaweel 2007a: Urbanization within
a dynamic environment: modeling Bronze Age communities in Upper Mesopotamia. American
Wilkinson, T.J., E. Peltenburg, A. McCarthy, E.B. Wilkinson, and M. Brown 2007b: Archaeology in the
Land of Carchemish: landscape surveys in the area of Jerablus Tahtani, 2006. Levant 39, 213–47.
Wilkinson, T.J. and D.J. Tucker 1995: Settlement development in the North Jazira, Iraq: a study of the
archaeological landscape. Iraq Archaeological Reports, 3. Baghdad.
Wilkinson, T.J., J.A. Ur, and J. Casana 2004: From nucleation to dispersal: trends in settlement pattern
in the Northern Fertile Crescent. In S.E. Alcock and J.F. Cherry (eds.): Side-by-side survey: comparative
regional studies in the Mediterranean world. Oxford, 189–205.
Winterhalder, B. 1980: Environmental analysis in human evolution and adaptation research. Human
Ecology 8/2, 135–70.
—. 1994: Concepts in historical ecology. The view from evolutionary ecology. In C.L. Crumley (ed.):
Historical ecology. Cultural knowledge and changing landscapes. School of American Research Advanced
Seminar Series. Santa Fe, 17–41.
Winterhalder, B., F. Lu, and B. Tucker 1999: Risk-sensitive adaptive tactics: models and evidence from
subsistence studies in biology and anthropology. Journal of Archaeological Research 7/4, 301–48.
Wirth, E. 1971: Syrien. Eine geographische Landeskunde. Wissenschaftliche Länderkunden, 4/5. Darmstadt.
Wolfart, R. 1967: Geologie von Syrien und dem Libanon. Beiträge zur Regionalen Geologie der Erde, 6.
Berlin.
Wossink, A. forthcoming: Climate, history, and demography: a case-study from the Balikh Valley,
Syria. In: Regards croisés sur l’étude archéologique des paysages anciens. Nouvelles recherches dans le bassin
méditerranéen, en Asie Centrale et au Proche et au Moyen Orient. Lyon.
174
Bibliography
Yalçıklı, D. and V.M. Tekinalp 2002: Mezraa Höyük excavations, 2000. In N. Tuna, J. Öztürk, and
J. Velibeyoğlu (eds.): Salvage Project of the Archaeological Heritage of the Ilısu and Carchemish Dam
Reservoirs. Activities in 2000. Ankara, 188–210.
Yardimci, N. 1993: Excavations, surveys and restoration works at Harran. In M. Frangipane, H.
Hauptmann, M. Liverani, P. Matthiae, and M.J. Mellink (eds.): Between the rivers and over the mountains:
Archaeologica anatolica et mesopotamica Alba Palmieri dedicata. Rome, 437–49.
Yoffee, N. 1988: The collapse of ancient Mesopotamian states and civilization. In G.L. Cowgill and N.
Yoffee (eds.): The collapse of ancient states and civilizations. Tucson, 44–68.
—. 1995: Political economy in Early Mesopotamian states. Annual Review of Anthropology 24, 281–311.
Zeder, M.A. 1994: Of kings and shepherds: specialized animal economy in Ur III Mesopotamia. In G.J.
Stein and M.S. Rothman (eds.): Chiefdoms and early states in the Near East: the organizational dynamics of
complexity. Monographs in World Archaeology, 18. Madison, 175–91.
—. 1998: Environment, economy and subsistence in northern Mesopotamia. In M. Fortin and O.
Aurenche (eds.): Espace naturel, espace habité en Syrie du Nord (10e–2e millénaires av. J.-C.). Bulletin of
the Canadian Society for Mesopotamian Studies, 33. Quebec, 55–67.
Zettler, R.L. 2003: Reconstructing the world of ancient Mesopotamia: divided beginnings and holistic
history. Journal of the Economic and Social History of the Orient 46/1, 3–45.
Zhang, D.D., P. Brecke, H.F. Lee, Y.-Q. He, and J. Zhang 2007: Global climate change, war, and
population decline in recent human history. Proceedings of the National Academy of Sciences 104/49,
19214–19.
175
Samenvatting
De uitdaging van klimaatverandering. Competitie en samenwerking tussen landbouwen pastoralistische gemeenschappen in noord-Mesopotamië (c. 3000–1600 v. Chr.)
Klimaatverandering heeft door de geschiedenis heen altijd een belangrijke invloed gehad op menselijk gedrag. Deze studie probeert inzicht te verschaffen in de complexe verbanden tussen dat
gedrag en klimaatverandering. De nadruk ligt daarbij op het effect van klimaatverandering op
de sociale interactie tussen menselijke gemeenschappen van verschillend karakter. Deze studie
betoogt dat deze interactie zich langs een continuüm zal ontwikkelen met competitie aan de ene
kant, en samenwerking aan de andere kant. De keuze voor een strategie wordt bepaald door de
mate waarin een strategie voordeel biedt wanneer een gemeenschap met klimaatverandering geconfronteerd wordt. Dit model wordt toegepast op de samenlevingen van noord-Mesopotamië
gedurende het derde en vroege tweede millennium voor Christus (3000–1600 v. Chr.), wanneer
zich een verschuiving naar een droger klimaat voordoet. Binnen dit geografische, chronologische
en klimatologische kader kijkt deze studie naar veranderingen in nederzettingspatronen als een
maat voor competitie tussen sedentaire landbouwgemeenschappen, en naar de ontwikkeling van
de Amoritische identiteit als een indicatie voor samenwerking tussen sedentaire en meer mobiele
pastoralistische gemeenschappen.
Hoofdstuk 1 bespreekt de wetenschappelijke context en de vraagstelling. Dit onderzoek maakt
deel uit van het multidisciplinaire project Settling the steppe. The Archaeology of changing societies in SyroPalestinian drylands during the Bronze and Iron Ages. Dit project wil inzicht verschaffen in de variëteit
en stabiliteit van menselijke bewoning in marginale gebieden in het Nabije Oosten. Binnen dit kader onderzoekt deze studie het effect van klimaatverandering op de organisatie en stabiliteit van
sedentaire gemeenschappen in noord-Mesopotamië tussen 3000–1600 v. Chr. Er wordt geconstateerd dat veel studies naar menselijke reacties op klimaatverandering sterk klimaat-deterministisch
zijn, zonder aandacht voor de mogelijke variëteit van menselijke reacties. De onderzoeksvraag is
daarom als volgt geformuleerd: hoe ontwikkelden zich de sociale relaties tussen sedentaire groepen
onderling, en tussen sedentaire en (semi-)nomadische groepen als reactie op klimaatverandering in
noord-Mesopotamië gedurende het derde en vroege tweede millennium voor Christus?
Hoofdstuk 2 bespreekt de geografische, historische en klimatologische gegevenheden van
noord-Mesopotamië. Het onderzoeksgebied beslaat delen van zuidoost-Turkije, noordoost-Syrië
en noordwest-Irak. Dit gebied is grotendeels vlak en wordt doorsneden door twee grote rivieren,
de Eufraat en de Tigris, en enkele kleinere. Noord-Mesopotamië kent een landklimaat met warme,
droge zomers en gematigde, natte winters. De hoeveelheid jaarlijkse neerslag varieert van minder
dan 200 mm. in het zuiden van het onderzoeksgebied tot meer dan 400 mm. in het noorden en
noordoosten. Dit betekent dat regenlandbouw in een aanzienlijk gedeelte van het onderzoeksgebied met grote risico’s verbonden is. Paleoklimatologisch onderzoek laat zien dat het klimaat van
noord-Mesopotamië gedurende het vroege derde millennium voor Christus gunstiger was dan
tegenwoordig. Studie van geologische profielen in noord-Mesopotamië heeft laten zien dat de
overgang naar een droger klimaat in de tweede helft van het derde millennium voor Christus moet
hebben plaatsgevonden. Klimaatreconstructies elders in het Nabije Oosten bevestigen dit beeld.
Noord-Mesopotamië bestond gedurende deze periode uit een verzameling van min of meer onafhankelijke lokale (stad)staten. Grotere regionale staten konden zich niet lang handhaven, en ook
externe machten, zoals Ebla, Akkad, en de Ur III staat, lijken vaak niet meer dan indirecte controle
of invloed uitgeoefend te hebben.
Hoofdstuk 3 bespreekt het theoretische kader van deze studie. Op basis van antropologische
literatuur over de reactie van (pre-)moderne menselijke samenlevingen op klimaatveranderingen
stelt dit hoofdstuk een model voor dat aangeeft welke vormen van sociale interactie verwacht mo-
177
gen worden als gevolg van klimaatveranderingen. Dit model verwacht enerzijds dat territorialiteit
en competitie zullen toenemen tussen groepen die afhankelijk zijn van dezelfde natuurlijke bronnen, wanneer deze bronnen schaarser worden als gevolg van klimaatverandering. Deze verwachting zal getoetst worden aan de hand van archeologische gegevens over nederzettingssystemen en
wordt verder uitgewerkt in Hoofdstukken 4 en 5. Het model verwacht anderzijds dat handel, uitwisseling en sociale cohesie zullen toenemen tussen groepen die afhankelijk zijn van verschillende
natuurlijke bronnen. Deze verwachting zal getoetst worden aan de hand van paleozoölogische en
historische gegevens en wordt verder uitgewerkt in Hoofdstukken 6 en 7.
Hoofdstuk 4 behandelt de analyse van nederzettingssystemen aan de hand van gegevens van
archeologische surveys. Er wordt een model besproken om gelijktijdigheid van nederzettingen te
simuleren, en archeologische en etnografische data worden gecombineerd om het aantal inwoners
per nederzetting te berekenen. Dit hoofdstuk toont aan dat etnografische data niet zonder meer
bruikbaar zijn voor archeologische case-studies en stelt een alternatief voor.
Hoofdstuk 5 bespreekt de resultaten van drie archeologische surveys die binnen het studiegebied vallen en vult deze resultaten aan met die van recente opgravingen. Deze surveys zijn de
Birecik-Euphrates Dam Survey (B-EDS), de Balikh Survey (BS) en de North Jazira Survey (NJS).
De methode zoals voorgesteld in Hoofdstuk 4 wordt toegepast op deze resultaten om tot een
evenwichtige reconstructie te komen van het aantal gelijktijdige nederzettingen en het aantal mensen per regio. Deze reconstructie laat zien dat de demografische en nederzettingsontwikkelingen
in ieder gebied zeer verschillend waren, en niet eenduidig verklaard kunnen worden. Daarnaast
worden de resultaten van omliggende surveys behandeld, en worden historische teksten besproken
die helpen om de resultaten van de analyse in context te plaatsen.
Hoofdstuk 6 behandelt het paleozoölogisch en historisch materiaal om de pastoralistische economie van noord-Mesopotamië gedurende het derde en vroege tweede millennium v. Chr. te reconstrueren. Deze analyse laat zien dat er gedurende deze periode een ontwikkeling plaatsvond
naar specialisatie van schaap/geit pastoralisme, waarbij ook de mate van mobiliteit/nomadisme
toenam. Het resultaat van deze ontwikkeling is duidelijk zichtbaar in het tekstmateriaal van Mari.
Hoofdstuk 7 behandelt de beschikbare historische bronnen over de Amorieten. De analyse
van dit materiaal laat zien dat deze benaming een etnische identiteit aanduidt. In tegenstelling
tot wat vroeger vaak verondersteld werd, waren de Amorieten reeds ver voor de Ur III periode
in Mesopotamië aanwezig, en werd deze benaming niet enkel door nomadische pastoralisten gebruikt. In tegendeel, deze etnische identiteit functioneerde als een identiteit welke verschillende
economische groepen met elkaar verbond en hen in staat stelde stevige sociale en economische
banden met elkaar aan te gaan.
Hoofdstuk 8 brengt de data uit Hoofdstukken 4–7 samen en analyseert ze in het licht van de
hypotheses uit Hoofdstuk 3. Er wordt geconstateerd dat de resultaten van de drie geanalyseerde
archeologische surveys maar deels in overeenstemming zijn met de hypothese dat competitie tussen sedentaire landbouwgemeenschappen zal toenemen als reactie op klimaatverandering kan geconstateerd worden. In de B-EDS regio lijkt de groei van Carchemish weliswaar aan klimaatverandering gekoppeld te kunnen worden, maar wordt deze groei niet door competitie gestimuleerd,
maar door Carchemish’ gunstige positie voor de handel tussen verschillende landschapzones. In
de BS regio laten de organisatie van het nederzettingsysteem, de constructie van stadsmuren en
historische bronnen een toename in competitie tussen nederzettingen zien voor de periodes waarin
de bevolking een maximale grootte bereikte. Hoewel deze reconstructie de hypothese bevestigt,
lijkt klimaatverandering hier niet de primaire oorzaak te zijn geweest, maar slechts een secundaire
factor die een proces versterkte dat al door normale bevolkingsgroei in gang was gezet. In de NJS
regio is evenmin een directe correlatie zichtbaar tussen klimaatverandering en een toename in de
competitie tussen sedentaire gemeenschappen. Het lijkt erop dat de dominante positie van Tell alHawa in het nederzettingsysteem, opgebouwd in het vroege derde millennium, het ontstaan van
andere grote nederzettingen verhinderde. Er wordt tevens geconstateerd dat de gelijktijdige ontwikkeling naar gespecialiseerd (semi-)nomadisch pastoralisme, en de opkomst van de Amorieten
gekoppeld kan worden aan gelijktijdige klimaatverandering en gezien kan worden als de vorming
van een identiteit binnen welke verschillende economische groepen konden samenwerken. De opkomst van Amoritische staten moet dan ook niet gezien worden als een infiltratie van nomadische
178
Samenvatting
stammen, maar als de reflectie van een proces van Amoritisering waarbij heersers de Amoritische
etniciteit adopteerden. De aanleiding hiertoe werd gevormd door de voordelen die de Amoritische
identiteit aan haar dragers verleende; namelijk de toegang tot een handelsnetwerk tussen pastoralistische en landbouwgemeenschappen waardoor de effecten van klimaatverandering gemitigeerd
konden worden.
179
Dankwoord
Op deze plaats wil ik graag collectief de collega’s bedanken die een bijdrage aan de totstandkoming van dit proefschrift geleverd hebben middels het kritisch lezen van (delen van) dit proefschrift,
discussies over de inhoud en richting van dit onderzoek, het toegang verlenen tot bronnen en artikelen waar ik zonder hen geen toegang toe gehad zou hebben en het verhelpen van allerhande
computer- en bureaugerelateerde problemen. Daarnaast wil ik graag iedereen bedanken die op
minder directe, maar niet minder belangrijke wijze aan dit proefschrift heeft bijgedragen door te
laten zien dat het leven in een ivoren toren zo slecht nog niet is, mij deelgenoot te maken van de
wondere wereld van de archeologie en anatomie van de Neanderthaler, de dagelijkse lunch tot een
plezierig medium te maken voor discussies over de universiteit in het algemeen en de faculteit in
het bijzonder en wat verder maar ter tafel kwam, de nodige afleiding van de wetenschap te verzorgen middels een biertje op zijn tijd, mijn paranimfen te willen zijn, en altijd interesse te tonen en
de nodige morele steun te verlenen.
181
Curriculum vitae
Arne Wossink werd geboren op 5 mei 1978 te Hilversum. In 1996 behaalde hij zijn VWO-diploma
aan de Minkema Scholengemeenschap te Woerden. Daarna begon hij aan de Universiteit Leiden
met de studie Archeologie, die hij in 2003 afrondde met als specialisatie Archeologie van het Nabije
Oosten. Aan het eind van zijn studie en tot korte tijd daarna was hij werkzaam bij een boekenantiquariaat in Leiden. In 2004 begon hij aan de Faculteit der Archeologie van de Universiteit Leiden
aan het promotieonderzoek waarvan dit boek het resultaat is.
183

Full text - Universiteit Leiden

Transcription

Similar documents

Bibliography and abbreviations